Biologie, Logique Et Metaphysiq - Devereux.pdf

all C A belongs to all B, but B need not be a primitive universal which is counterpredicable with. A . B belongs to all C (e.g. a very specific class), so A belongs to all C. An example might be: A: being broad leafed, B being a green leaf shedder, C being a vine. In this case, being broad leafed cannot explain why all vines shed their leaves since the A-term extends further than vines to other trees. Thus, A may be an explanatory term, provided that it is not explanatory of why B l/J all C, but rather of why B

knowledge required to know non-incidentally that a phenomenon or kind exists:

150

'

I

.!

(7) In this construal, the A-term is explanatory, but not of B cl> all C. Hence there is no countepredication. The final set of cases discussed in this chapter (78b13 ff) then will be t.he only ones where a cause is not mentioned at all. For a contrasting view, see Barnes: Commentary on Posterior Analylics p. 150. (8) This is important in considering the problemala analysed in Post. An. 1 1 . 14. See note below.

i

151

"As to whether it is, sometimes we grasp this incidentally, and sometimes through grasping something of the thing itself - e.g. with respect to thunder, that it is a given type of noise in the clouds, with respect to eclipse, that it is a given type of depriv ation of light, with respect to man that he is a certain type of animal..." (93a21-4)

This passage has many difficulties.• But it is clear that in some cases grasp of the genus (animal) and differentia (a certain type of...) is, at least, necessary to grasp the existence of man. However, this remark is only illuminating if we already know how the relevant genera and differentiae are fixed and how they are understood to be so fixed. For if we knew this, we might at least know something of how the existence of kinds such as man is established. Unfortunately in this passage, Aristotle simply assumes that the relevant genera and species have been fixed, and does not say how this has been done. In Post. An. I l . 13 also (96b25-97a6, 97a23-28) Aristotle seems to assume that the genera and appropriate differentiae have already been fixed, and then focuses his attention on their proper ordering rn Similarly in I l . 1 4, Aristotle's method is this: grasp the genus (e.g. animal) and see whatever belongs to that genus (98a3-4). This allows for the problem to be correctly formulated as (e.g.): Why does F (something which belongs to all animals) belong to all animals? The relevant next step is to note whatever follows from the next of the divisions down: e.g. bird (G). For this allows for a further problem: Why does G belong to all birds? This method forces upon us the correct formulation of problems. ' Why does F belong lo birds?' is not a proper problem since being F belongs to birds, only because birds are animals. Thus it is clear that the point of introducing this structure is to ask questions at the right level for apodeiclic proof: to see what belongs to what non-incidentally, and what incidentally qua subspecies of a genus to which it belongs essentially. (9) I have discussed some of them in Aristotle on Meaning and Natural Kinds. One major problem is whether 'grasping something of the thing itself' is a necessary or sufficient condition for knowing that the phenomenon exists non-incidentally. I intend to discuss this issue - which requires a detailed analysis of the lines 93a25-35 - in detail elsewhere. (10) These passages require further detailed analysis. In them (e.g. 96b31-5), it seems that Aristotle assumes that at each level differentiae will divide genera (or sub­ genera) to form two further genuine kinds by pointing to features which are temselves explained by the essence of the new kind. But this assumption rests on a view about what type of properties are 'differentiating' ones, which is not at all clarified in these passages.

152

NATURAL KINDS AND NATURAL HISTORY

The basis for this method rests on a prior grasp of what counts as a genus and its subdivisions in appropriate order (98a2-4). Of such a genus with its subdivisions thus ordered one can ask: does F belong essentially to genus or subdivision, as the correct basis for setting up the problem of " Why? " . It is not that one is using the fact that F's belong to all A's to fix a genus or sub-species. It is rather the other way about: one begins with a grasp of genus and sub-species and then lines the predicates up to them essentially. One is using a set order of_genus and sub-species (divisions) to determine which predicates belong to what essentially. One is not in this context using the issue of which predicate belongs to what essentially to fix genus and sub-species. This structure seems particularly clear in 98al3-19. Here Aristotle writes:

In this way I l . 1 3 and I l . 1 4 are closely connected. Aristotle says that division is usdul for pursuit of what a thing is (96b26-28), but the uses he envisages presuppose that one already grasps the genus (96b3637), within which one marks out further distinctions in a systematic way. While division may be of help in introducing the latter distinctions in an orderly way, Aristotle does not suggest that division is useful in fixing what is essential. For he writes:

(a) establish Vx [Gx-+ Fx], where 'Fx' means x is horned; (b) establish Vx [Fx-+ Hx], where 'llx' means x has manyplies; and finally (c) What animals have F? An answer to (c) might be: type A (goats). But while this requires that Vx[Ax-+ Fx], it does not require that V[Fx-+Ax]. Goats might be one of the kinds which are F and not the class of anirnals all and only which have F. (e.g. 99a32-5). If so, there is no essential use of commensurate universals or counterpredication in this passage. Similarly in 99b5 ff, while all birds may have a given property, possession of this property need not be confined to birds (e.g. possession of a gut of a given kind) provided that not all animals possess a gut of this kind (cf. Hisl. Anim. Il.17: cf. 508b1417). It is not required that the universals at this stage belong exclusively to birds (cf.

99a32-5), but some could apply to (e.g.) fish as well. I doubt whether the correct formulation of problems in I l . 1 4 requires counterpredication or commensurate universals. These features enter at a further stage {cf. 99a32-6) where one reaches the end of the explanatory project. Thus, 'why do all birds have a gut of a given kind? might be an acceptable problem, even if all fish also had such a gut.

In this passage, Aristotle (it seems) asks three question: (i) what properties does (e.g.) having horns follow? (ii) what properties follow this property? (iii) what objects have horns?

I

I' I '

I

I ! '! ;I

"To establish a definition through divisions, one must aim for three things - getting what is predicated in what the thing is, ordering them as first or second, and ensuring that these are all there are." (97a24-26)

The first of these involves establishing 'things through the kind': i.e. assembling those properties which are essential properties of the kind and this seems independent of the method of division. That enters at a further stage provided that all the definitional elements already introduced can be uniquely arranged in a descending order. But the process of division here is just a way of arranging definitional predicates. It presupposes (and does not supply) an account of what the definitional predicates are. Thus, both II.13 and 1 1. 1 4 depend on an already-fixed account of what genus and defining properties are. What is to count as a genus is also presupposed in Aristotle's critical discussion of division in the Paris of Animals. I. His criticisms hinge on two points. (a) Division by itself does not show what is properly taken as a genus. If we divide in certain ways, then we mistakenly cut across genuine natural kinds (642b10-20), divide by essential accidents not by things that are in the essence (643a27-32) or divide by the wrong things/features (643a35-b6). In these cases division leads to inaccurate results because it does not rest on a genuine account of what naturally belong to the genus or essence of a phenomena or kind in question. But if so, an account of the genus and essential properties is presupposed, and not established by division. (b) Correct division must be by multiple, simultaneous differentiae (642b21 , 643b9), where relevant differentiae are applied to the genus simultaneously. Both of these points rest on the claim (already familiar from Post. An. I l . 13) that division rests on genera already established. It does not establish what the genera are. Nor does it by itself determine which the species, or relevant differentiae are-.

'we must extract a common property and ask what (sc. qualities) it follows, and what qualities follow it : e.g. having a manyplies and lacking incisors follows having horns. Again we should enquire to what animals having horns belongs. For it will be clear why having a manyplies belongs to them: because they have horns.'

I I

153

D. CHARLES

(iii) should be different from (i) - as signalled by 'Again . . .' in the text. If so, (iii) should be concerned with types of animal, not properties [as (i) must be if it is to balance (ii)]. In, (iii), Aristotle is assuming that there are genera or species of animal which are horned, but which lack one common name (like 'bird': 98a7). These genera or species might be described by a phrase (wingless quadruped), or by several names. But they will not be picked out only by the phrase 'horned animals'. For in that case, the answer to (iii) would be trivial. There is no evidence for taking ' being horned' as a genus ia this passage. Aristotle is depending on genera and species marked out by other means in this context also.11 (11) Aristotle is envisaging three stages:

I '

154

D. CHARLES

NATURAL KINDS AND NATURAL HISTORY

If we knew, in given cases, what the genera and species were, we would have at least the basis for establishing the existence of a given kind. But Aristotle seems to presuppose these answers in both the Analytics and in this section of the Parts of Animals. We are no wiser as to how to establish the existence of kinds at the end of them. So there are two major questions left unanswered in these discussions: (a) What is required to establish the existence of a kind? (b) How are kinds to be differentiated - some as genera,· some as species? These questions are, of course, related. To establish the existence of a kind is to establish the existence of either a genus or a species. To provide successful differentiations into genuine kinds is to establish their existence. However, in the highly abstract discussion of the Analyfics, these issues are mooted rather than resolved. If they are tackled at all, it is in Aristotle's detailed discussions in the Biological Works. To these we should now turn.

far in determining under what conditions species are not far removed from one another. Aristotle does indeed say that: 'it is practically by resemblance of the shape of their parts, or of their whole

J I . THE INPUT OF THE BIOLOGICAL WORKS In Parts of Animals 1.4., Aristotle is concerned with deciding which of two methods to follow in describing the properties of genera and species. One method consists in · studying each indivisible species · separately (644a29-644bl). Aristotle rejects this on the basis that it is ' long-winded' , making one describe the same attribute time and time again as they are common attributes of many species. He continues: "So perhaps the right procedure is this: (a) so far as concerns the attrib htes of those groups (genera) which have been correctly marked off by comri:ton usage - groups which possess one common nature apiece and contain in themselves species not far removed from one another (I mean birds, fishes and any other group which though it lacks a name yet contains species generically similar) - to describe the common attributes of the group all together; and (b) with regard to those animals which are not covered by this, to describe

i

the attributes of each by themselves; e.g. those of man, and of any other such species." (644bl-9)

The passage contains the following thought. Certain genera have heen correctly marked out by common usage: those that (i) possess one common nature, and (ii) contain species not far removed from one another. It appears that - in this passage - one genus will be established if and only if it contains (i) a distinctive common nature and (ii) species not far removed from one another. This passage, however, is also elusive. It does not say clearly what constitutes one distinctive common nature. Nor does it advance very

155

body, that the groups are marked off from one another - as groups of birds, fishes, cephalopods, testacea' (644b 10-12),

and further notes that within these groups (genera), the parts differ only by 'the more or the less' - e.g. by being larger or smaller, softer or harder (644bl3-15: cl. 644a l5-25). But these remarks also are under­ specified. Aristotle does not make it clear whether it is difference in shape of parts that constitutes (in virtually all cases) difference in nature,. or whether different natures marked out on different criteria do - in general - in fact have differently shaped parts. At the end of Part. Anim. 1.5 Aristotle notes that the parts of the body 'are for the sake of some action' (645b l6-l 7) and exist for the sake of their natural function (hi 9-20). Relevant actions include generation, growth, copulation, waking, sleep, locomotion (ibid. b35 ff). Is it these functions that define similarity and difference in parts of animals independently of the shape of the parts, so that two parts in different animals may he the same even though they have differing shapes provided that they play the same functional role in similarly organised structures? Or is it shape that is basic, so that major differences in shape constitute difference in parts irrespective of similarity in functional roles? The Parts of Animals raises, rather than resolves these issues rn Aristotle does not say which of these features are constitutive of having 'a common nature' - or which are 'essential attributes ' of a given group of animals (645bl-3). At the beginning of Part. Anim. I i . Aristotle refers us back to the History of Animals (646a8-10). And it is there that Aristotle focuses most insistently on these issues. Thus, in Hist. Anim. 49la5-10, he writes: "Our goal is to establish first the different that exist and the properties that belong to them all."

which he takes to be the task of the historia." This suggests that in these writings he is engaged in marking out the genera which are, in reality, different from each other (as in the immediately preceding (12) Similar issues infect Aristotle's treatment of the 'the more and the less'. Do organisms' parts differ by 'the more and the less' when they subserve similar functions but do so in a 'shape-wise' different way? Or do they differ in this way when they are 'shape­ wise' similar enough, even when they subserve somewhat different functions. The remarks in Parl. Anim. 1.4 leave these issues unresolved. (13) In 491a8-10, Aristotle is not merely trying to grasp any feature which distinguishes animals - for then why separate the existing differences from those which belong to each animal? This shows that in the Hist. Anim. he is engaging in a taxonomic investigation, even if he is not aiming at a complete taxonomy of all the species that there are.

156

NATURAL KINDS AND NATURAL HISTORY

section of Hist. Anim. 1.6) and then in grasping what properties belong to each member of a genus thus marked out.14 So we may address to these writings the following questions: (a) does Aristotle here succeed in marking out the notion of a common nature required to separate differing genera of animals? (b) does Aristotle succeed in marking out (in outline) the basis for species-differentiation within a genus? Positive answers to (a) and (b) would assist us in filling the lacunae spotted in the A nalytics: (i) what determines what the relevant genera and species are? (ii) what is involved in establishing the fact or existence or a genus or

- where the last four are bloodless. How did Aristotle arrive at this classification? Why did he exclude certain other candidates as genera or great genera? (490bl5-49la6) Aristotle's earlier discussion in Hist. Anim. 1 . 1-5 has laid the foundation for these claims. Birds have been marked out as feathered fliers (490a6-7, 12-13), fish are as gilled, finned and footless swimmers (489b22-6). Cetacea are also swimmers, but lack gills and instead have a blow-hole (489b2). Insects are fliers with membraneous wings (490a9l l ) and by contrast with other fliers are bloodless. The 'softies' (cephalopods) are swimmers, possessed of feet and fins (490al-3). The soft-shelled animals (cruslacea) are all capable of locomotion (487bl6-18 - whether by swimming or walkings), while many types of hard-shelled animals (testacea) are stationary (487bl5), and take in neither water nor air (487a25-6). Aristotle, it appears, has provided a justification for treating these genera as correctly marked out by common usage. This has been done by showing how these kinds exhibit genuine differences in the following respects: (a) type of locomotion: walking, flying, swimming; (b) method of breathing: with lungs, gills, blow-holes; (c) method of eating: where the animals find their food (487a22-7). In some of these cases, modes of locomotion are the basic ingredients. Fliers are divided according to their method for doing so (feathers, membraneous wings). In others, modes of breathing are basic (gills, blow-holes etc.). But each of these kinds is marked out using basic divisions within the life functions (locomotion, breathing, eating etc.). It appears that it is these a ffections and actions that provide the basis for Aristotle's classification at this point (cf. Part. An. 645b35 ff). To have a common nature is to have a distinctive way of fulfilling at least one of the basic life functions. This claim becomes clearer when Aristotle considers cases of groupings which fail to yield great genera. Wingless quadruped is briefly canvassed as a possibility in 490b19-20, but not used elsewhere as a genus. By contrast, viviparous quadruped and oviparous quadrupeds are introduced and retained as relevant genera, although neither are great genera. Why is this? Aristotle notes that wingless quadrupeds would divide into the viviparous and oviparous (490b20-l). These are nameless groups, but so are the immediate sub-divisions within the genera of viviparous quadruped (490b32-3). The explanation seems rather to be that if one had a genus such as wingless quadruped, the species within this genus would not be very similar (Part. An. 644b4-5) in a certain crucial respect: method of reproduction. And from this difference, other important differences follow - with respect to hair and horns (490b21 -3).

species?

I

I

157

D. CHARLES

For if Aristotle's use of the idea of common nature is important for his investigations, this may provide his basis for establishing the existence of genera and species. Thus, to know that a genus or species exists is to know that it possesses a common nature of some type (possibly so far unspecified). But to know this is to know that there is a cause (viz. its common nature) which explains why (e.g.) gold or fish exists, or possesses certain properties - even if we do not know what the cause is. To know that there is a nature guarantees the existence of a kind even before we grasp what that nature is. It points to the existence of the unifying principle or nature required if there. to be a genus or species whose existence is established at the second stage of the enqmry. My main aim is to see how far the Hisloria A nimalium provides insight into the dark places left in the A nalytics discussion. But this can only be achieved if the reading to be offered of the Hisloria makes sense of Aristotle's argumentative structure and deployment of data in this work. So my subsidary aim is to suggest an outline for upder­ standing Aristotle's pro jet in certain central parts of the Hisloria.

I I I . H ISTORIA ANIMALIUM In Hist. Anim. 1.6. Aristotle lists the seven main groups (genera) of animals: birds, fishes, cetacea (all of these are blooded), hard-shelled fish, soft-shelled animals, 'softies' (types of calamary) and insects (490b7-14) (14) See Part. Anim. 645bl-3 'our task is to distinguish the properties (sumbebekota) which belong to each group. I mean the essential attributes which belong to all the animals .. .' It seems that in 491 a10-11, the properties (sumbebekota) should be of this type - viz. the object of possible demonstrations (cf. Pr. An. 46a24-7), and not merely accidental properties.

158

NA"TURAL KINDS AND NATURAL HISTORY

In this example, Aristotle is relying on two ideas. If one has a genus, the species which comprise it should perform basic life functions in a basically similar way (e.g. reproduction). Further difference in mode of reproduction is sufficient to show that there are different genera in the case of quadrupeds. Both show that Aristotle is concerned to secure his favoured genera on the basis of a unified common nature - in which shared ways of performing certain basic functions are of central importance (locomotion and reproduction, in this case).15 In the case of a unified genus there is a distinctive way of moving, repioducing, feeding and breathing. Difference with respect of one of these life functions undermines the unity of the genus. The idea of a common nalure rests on the thought of one organized collection of methods of moving, reproducing, feeding and breathing. A powerful theory of what constitutes a common nature is at work in selecting the great genera in 1.6, and in ruling out other possible cases.

means of the ways in which the animals reproduce (internally, externally, viviparous etc.). He only goes further in 507a34 by invoking a further sub-division within the class of viviparous quadru­ peds: those that have horns but lack teeth in both jaws. But this further demarcation is forced on him since at this point he is focusing on differences in types of stomach. Not all of Aristotle's X-places are filled by types of animals marked out using features relevant to generation and locomotion in this section. Other cases are:

The theory is at work elsewhere in the Hisloria. It is a recurrent feature of this work that Aristotle attempts to line up kinds of animals with given features in structures of the following form: 'As many as are X, some/all/none are y.' 16

In discussing the instrumental parts of blooded animals, Aristotle take the following cases as standing in the X-place (505b25-523a27). viviparous quadrupeds (505b32, 506b26); oviparous quadrupeds (505b35, 506b26); viviparous quadrupeds with horns and do not have teeth in both jaws (507a34); footless not internally viviparous (509b6); footed viviparous (510al3-14, 517b2-3); . footless animals that are externally viviparous though internally oviparous (51la4, 52l b25); footed ovipara (5 l 7b4-5); blooded animals that are internally and externally viviparous (520b27, 52lb22).

' ' I '

!

. 'i, ,I1' . , : .. II, i

In these cases, Aristotle 1s again usmg features relevant to locomotion and reproduction as the basis for dividing up types of animals. Subsequently when he divides the kinds furlher than the genera marked out in Hisl. An. 1.6. he frequently does so by dividing by (15) Why are serpents not a 'great genus'? Is it because this 'genus' too cuts across animals with different means of reproduction (oviparous/viviparous: 490b24-27), where the species are too different from each other? Aristotle leaves this point in the air in 1.6, and proceeds later to treat serpents in certain respects like other oviparous land-animals {508a9-12), although he notes that serpents are divided as to their habitat {505b6-10). (16) This point is clearly and correctly emphasized in Jim Lennox's paper: Divide and Explain. The Posterior Analylics In Practice in Philosophical Issues in Aristotle's Biology (ed.) Gotthelf and Lennox C.U.P. 1987.

,I

I

159

D. CHARLES

blooded animals (506a6-14); animals that take in air (506a2); animals that have kidneys and bladders (506b32).

But the latter two examples are dependent on the subject-matter in the passage under discussion: oesophagus and windpipe in the first, kidney and bladder in the second. 'Blooded' is somewhat different and is used quite widely; but this seems to refer back to the first three great genera (fish, birds, cetacea) mentioned in Hist. Anim. 1.6, and to the other cases of blooded animals later adduced (e.g. man: cf. 505b29-32). The basis for differentiation remains organized natures with distinctive means of reproduction and locomotion in these passages." It seems that throughout these sections Aristotle's demar­ cations rest on classifications dependent, in the main, on differing ways of fulfilling the basic functions of reproducing, moving, breathing, and growth (Part. An. 645b33-5). It is these which constitute the genera - whether great or less - marked out by differing types of organization for fulfilling these functions. Aristotle's intuitions about common nature depend on his view of the basicness of locomotion, breathing and reproduction in the study of animals. On this foundation, he has good grounds for marking out the great genera of 1.6 - in terms of differing ways of moving, breathing or reproducing. Once he has marked out a common nature in this way, he is able to mark out differing sub-kinds in a structured way. Fish are gilled, finned swimmers (489b24-5). He is then able to use these differentiae of fish to divide the genus further: (a) 4 - finned us 2 - finned us no-finned: 504b27-35; (b) covered us non-covered gills (505al-2); (c) single us double gills (505a8-IO); (d) few us many gills (505a 10-12);

and thus to separate several major species within the genus of fish such as selachia (504b36), muraena (504b34, cl. 489b24-6), eels (504b27-30 cf. (17) A similar story is present in Aristotle's discussion of miscellaneous parts {533a 1558b6). There, too, the X-places are filled by e.g. four-footed ovipara (536a4), footed, blooded, and non-ovipara {538a21), man, footed, vivipara, blooded ovipara {533al), blooded oviparous quadrupeds (557b32).

160

I

I

h!

D. CHARLES

N.A:TURAL KINDS AND NATURAL HISTORY

489b24-6), as well as other smaller species (such as the mullet, parrot­ wrasse, perch, rainbow-wrasse, carp, dog-fish, sword-fish, tape-fish, fishing-frog - all discussed in Hist. Anim. I l . 1 3). All fish have other features in common also (in Aristotle's· account). All are blooded (505b2-3), all have a bladder (506b4) although this may be differently placed in different fish (Hist. Anim. I l .15). All have a gut of a given type - a feature they share with birds (508b l4), although they differ from one another in the number they possess (508bI5-25). Aristotle uses these differentiae to mark out varieties of fish (at a fairly specific level: e.g. angel-fish, skate, pipe-fish, goby, burbet etc.). But these are all varieties of fish because they carry out similar functions (nutrition of a given type) in similar ways, and all possess the differentiae of fish (they are either gilled or finned swimmers). Through employing these intuitions about 'common nature', Aristote is able to vindicate the popular assumption that fish are one genus of animal - with similarity in function and in structural features related to those functions. Against this background, he is able to mark out major differences between fish in other respects: some, the scaley fish, are oviparous, others viviparous (505b2-4) - for instance, the selachia. Indeed, he defines the selachia as footless creatures which have gills and are viviparous (51 l a5-7)18 Selachia have distinctive sexual differentiae (538a29), bone structure, and reproduce in a distinctive manner (540b5 ff, 564b 1 5 ff). Thus, Aristotle is relying on the importance of the soul-function of reproduction to distinguish lower­ level types of fish - where all are varieties of fish because all are blooded and either breathe or move in the way distinctive of fish. There is a determinate enough common nature of this type to unify the class. Indeed what it is to have a common nature is to have one organised mode of performing certain of these basic soul functions .. Once this common nature is established, Aristotle is able . to differentiate between fish in terms of how they perform other soul­ functions: generation ( I I I) perceiving, hearing, sleeping (IV), reprodu­ cing (V, VI), eating (VIII), migrating (VI I l . 1 2 ff), hibernation (VI I l . 15), flourishing (VI I 1 . 19), illness (VI I 1.20), adaptation to their environment (IX.37). O f these, procreation and eating are the most basic (in Aristotle's view): 'One part of the life of animals consists in actions concerning procreation; another in actions concerning food. For it is one these two activities that interests and life of all are fixed'.

I. I

(18) What happens to the fishing frog (505b2)? subsequent discussion.

It seems to remain submerged in the

161

And elsewhere he notes: 'The habits of animals are all concerned with breeding and rearing of young, or with obtaining food'. These habits are modified so as to suit cold and heat, and the variations of seasons'. (596b21-24)

In discussing these activities, Aristotle often begins his account by saying what all fish have in common: they all lack testacles (509b3, 25), none have breasts or milk (52l b25-6), all lack visible sense-organs (533a30), all sleep (537al), and all feed on spawn (59la6). But then he proceeds to note that differing kinds of fish perform these functions differently. The case of eating in Hist. Anim. V I I l .1-12 is instructive of Aristotle's method, and I will again focus on his treatment of fish rn In V I I I.2. Aristotle in general is concerned to note differences in eating patterns between the different types of fish he has already marked out on separate grounds such as: (a) (b) (c) (d) (e)

types of gill: grey mullet, scarus, selachians, dentex20; types of fin: conge�, muraena, bass, gilt-head, bass, eels21 ; types of gut or gall: amia, red mullets, cephalus, goby"; types of skin: tunny"; types of breeding or spawning pattern: sea-perch, saupe, physics, rock fish, mackerel, sargue, channa.24

In these distinctions, modes of breathing, moving, digesting and reproducing play the central role. Aristotle, in nearly all cases, is mapping distinctive ways of finding food on to types of fish which he has (19) In V I I I . 1-12, Ari.stotle is concerned to map differing ways of obtaining food on to his favoured genera (fish, birds, serpents and other scaly anim3is, viviparous, quadrupeds, and insects), and their sub-species. Thus, in the remainder of V I I I . 2., Aristotle operates with the following demarcations of aquatic animals: 590a l8: hard-shelled animals (leslacea) that are incapable of motion; 590bl ff: mobile shell-fish; 590bl0 ff: crustaceans; 590b30 ff: soft shelled animals; 591al ff: octopus; 59Ia6 ff: fish. He then discerns their differing ways of finding food - as was to be expected given their differing natures - and marks further differences within· the respective genera. (20) Grey mullet (59I a l 8 cf. 504b32), scarus (59lal4 cf. 489b26-7), selachians (59Ial0 cf. 489b24-6), dentex (59 la l l , 505al6). (21) Conger (59Ial0, cl. 489b26), muraena (59Ia l l cf. 489b24-6), bass (59Ial l , cf. 489b26) gilthead (59 I b l 0 cf. 489b26), eels (59lal8, b31, cf. 505a l6). (22) Amia (59Ial0, cf. 506bl3), ced mullet (59 I a l 2, cl. 508bl7) goby (59lbl4, cf. 508bl7). (23) Tunny (59l a l l , cf. 505a27). (24) Saupe (59Ial5, cf. 543a8), phycis (59lbl3 cf. 567b20), rock fish (59lbl4, cf. 543a5), sargue (591b19, cf. 543a7) channa (591a l l cf. 538a21), sea�perch (59 l a l l , cf. 543bl).

162

I

. i I

I

I

163

D. CHARLES

NATURAL KINDS AND NATURAL HISTORY

already distinguished in terms of other soul-functions. Thus, he is confirming his earlier classification by showing how these differences are related to other differences with respect to eating. The only three (clear) cases of fish species added in V I I I.2 are the dascyllus (59l a l 4) melanurus (59 l a l 5) and the peraeas (a24). All others have been already marked out in terms of their other soul-functions. Indeed this is the general pattern in Aristotle's discussion of migration, hibernation, flourishing and adaptation. In the overwhelming majority of C?Ses, the types of fish attributed different kinds of (e.g.) migratory pattern have been marked out using the basic distinctions already mentioned. While he is prepared to add four further types of fish to his list at this point (e.g. black bream, 598a l0; weaver, 598al2; braize 598al3; sea cuckoo (598a!4), the remaining twenty-four have already been distinguished on other grounds. In these cases Aristotle is separating distinct sub­ species of fish in virtue of their basic natures - modes of moving, breathing, digesting and reproducing, while occasionally underwriting further groups which differ in eating or migrating pattern. The theory of soul-function is at work both directing his enquiry and underwriting his vindication of popular claims about differing types of fish. In the cases so far considered the theory of soul-function is used to differentiate the genus of fish, and to mark out further differences within this genus. But elsewhere in Hist. Anim. V I I I . it is used to separate different types of aquatic animal. Some live and feed in water, take in water and emit it (e.g. fish), while others live and feed in water but take in air not water and breed away from water (e.g. otter, beaver, crocodile). (487al8-20) Some again obtain their food in water, but take in neither water nor air (e.g. shellfish, sea-anemones: cl. 487a23-25). In these cases, mode of breathing and location of food are used to distinguish different types of water-animal. Aristotle then links them to an extent by commenting that no water-animal which takes in sea­ water gets its food from land. (487b3-5) In this passage, breathing and mode of nutrition play a decisive role in separating out animals which share the same habitat, and in marking out different way,s of being aquatic. In his subsequent discussion of these animals (Hist. Anim. V I I I .2.), Aristotle is forced to further refinement in his treatment of the aquatic through consideration of the dolphin (589a33-b27, 590a l3-18). The dolphin is remarkable and a problematic case because it takes in and discharges water by his blow-hole, and yet also inhales air into his lungs (589b5-6). Thus, the dolphin takes in and emits both water and air, and hence could be classified as both aquatic and terrestrial (589b714). Faced with this problem, Aristotle moves to further define (prosdioristeon: 589bl3) the aquatic as follows: some aquatic animals take in water (via their gills) for the same purpose as respiration - to cool

their blood; others take in water for the different purpose of obtaining food; and hence do not use water to cool the blood but expel it via their blowhole (589bl5-18). These animals use air to cool their blood, and hence are aquatic in a different way from fish. Difference in breathing distinguishes different ways of being aquatic: one type of animal breathes by taking in water, the other by taking in air (590a l4)." Fish and dolphins, however, take their food from the water, and thus can be distinguished from animals which breathe like fish but obtain their food on dry land (the water-newt: 589b25-8). In these cases, Aristotle is using basic soul functions (breathing, eating and locomotion) to distinguish different ways of being aquatic. And this is what one would expect given that he is carrying through his investigation with a robust theory of what is to count as a common nature. When animals without a common nature are both characterisable as 'aquatic', this shows that the latter term needs further definition." This pattern of argument suggests that Aristotle in the Hist. Animalium is beginning with a set of beliefs about what possessing a common nature consists in: being an animal with an organized set of soul­ functions involving (typically) locomotion, breathing, reproducing and eating. To establish that a great genus exists involves establishing that there is an organized and distinctive set of soul-functions of this type, where one can see further species as differentiated (in large part) by their differing ways of performing the same set of soul-functions. To establish that a species of genus exists is to establish that there is a nature which performs the set of soul functions of the genus in a sufficiently distinctive manner.27 In both cases, one is grasping that there is a common nature involved, even if one has not fully grasped what that nature is: what all the precise soul-functions involved are. According to this view, Aristotle is not in the Hist. Anim. engaged in a wholly a priori investigation of the natural world. While his theory of soul-functions is drawn from his psychological works, it is an empirical matter to determine what the relevant genera or species are:

(25) In 590al4-15 Aristotle also adduces the mixture of the body (Krasis) as a way of distinguishing different types of aquatic animal. This appears to depend on the surrounding medium by which the animal is cooled and the food it takes in (Gen. Anim. 767a30 ff). If so, it is determined by features relevant to breathing and food, and so is derivative from these more fundamental categories. (26) This passage has interesting implications for the study of dualizers. It suggests that dolphins dualize only provisionally between being aquatic and being lerreslrial. They do not dualize when the definitions of being aquatic are further tightened (589b12 ff} . This point is discussed by Robert Parker in his review of G. E. R. Lloyd's: Science, Folklore and Ideology: Phronesis 1984 p. 184. See also Herbert Granger's: Scala Nalurae and the Continuity of Kinds. Phronesis 1985. (27) This is further clarified at stage ( I I I} : see below. ,

I

D. CHARLES

NATURAL KINDS AND NATURAL HISTORY

what types of organized soul-functions there are.28 But on the other hand his work is not wholly empiricist either. He is not fixing a genus merely by collecting a large group of counterpredicable properties, and then using division to establish the species. Rather the weight given to differences in basic soul-functions shows that he is relying on strong background assumptions about what features are relevant to establi­ shing genuine common natures. His enquiry is, in effect, the progressi­ vely more systematic elucidation of certain controlling concepts of this type which results empirically in differentiation into kinds and sub­ kinds.29 It is clear that Aristotle in the Bisi. Animalium is not aiming at a complete taxonomy of all animals - an exhaustive, hierarchic

classification. For his account (as has been frequently remarked) is incomplete in many respects. Nor will all the differentiations he makes turn out to be relevant to the classification of kinds or sub­ kinds. Rather what he is seeking to do (on this view) is to show the adequacy of his theory of soul-functions to mark out the differing species of animals - even though he has not himself completed the task itself. The Hisloria is a concealed argument for the centrality of these concepts in establishing the existence of genera and species - even if not all have been actually established by Aristotle's methods in this work. He is giving a blue-print for a successful taxonomy, rather than attempting to complete that taxonomy himself. But he is attempting far more than laying out all the relevant animal differentiae.•0

(28) Contrast I . During's claim in Aristotle's Method in Biology in Aristote et les problemes de methode (ed.) S. Mansion p. 218: 'even in his most advanced biological works his reasoning is based on a priori principles and book knowledge more than on observation.' And elsewhere 'It does not seem reasonable to characterize his method as empirical' (loc. cit.). (29) I take it that an empiricist reading of the Historia Animalium, which saw it as concerned with establishing the existence of kinds and species, would represent Aristotle's method as follows:

IV. STAGE I I AND STAGE I I I

164

iI I

,ii

stage (I): use counterpredication to establish genera as the locus of many such counterpredictions; stage (II): divide within such genera using further counterpredication and division to establish species, On this account, stages (I) and (I I) would be innocent of background theoretical assumptions about what constitutes a common nature. I have been arguing against this view by emphasizing the significance given to certain factors as constitutive of a common nature throughout Aristotle's discussions of genera and species (e.g. in the case of fish). Further, even where Aristotle could set up wider genera (e.g. the blooded in Hist. Anim. 1.6 or the lung-possessor: Part. Anim. 669b8-10) using counterpredication, he does not do so when the distinctions cut across his favoured genera with their distinctive common natures. His treatment of being a lung-possessor is especially interesting. Aristotle notes that because there is no common name which applies to those genera which have bloodless lungs, this fact appears in the specification of their essence. Thus, here Aristote is noting that there is no one genus of (bloodless) lung-possessors in contrast ' to the case of birds. Rather having lungs of a given type appears in the specification of different genera marked olit on other grounds. Counter-predication is not enough to establish a genus. What is needed to do this is the feature which all birds possess and lung-possessors do not: viz. a sufficiently rich common nature. Further, once genera have been set-up - such as fish which are all blooded (505b2-3), some are viviparous, some oviparous, some have an oesophagus (11.17), and all have a gut like that of birds (508b14-15). Counter-predication plays a very small role in establishing the genera themselves. It is one way amongst many of establishing that there is a common nature. Certain elements in Lennox's recent papers on these topics (Divide and Explain: see footnote 16; Between Data and Demonstration: the Analylics and the Hisloria Animalium: 1988. Science and Philosophy in Classical Greece) suggest that he advocates an empiricist reading of the llisloria A nimalium (subject to certain qualifications). Lennox's views, however, are subtle and evolving, and I hesitate to represent them as constituting the 'out-and-out' empiricist reading sketched at the beginning of this note.

165

The assumption o f a common nature a t stage I I embeds an explanatory notion at the pre-(scientific) explanatory stage, which serves two purposes. (a) It explains why what is grasped at the earliest stages (I and II) by way of genera and species is explicitly connected with stage I I I . For what i s discovered a t stage I I I explains the basis for a common nalure (e.g. shared teleological goals and shared type of material basis: degree of heat). It is no accident that the genera and species marked out at stage I I match those which science explains since the latter will fill out and further explicate the guiding conception of a shared nature already implicit at stage I I . If this notion had not been invoked at the second stage, and the genera and species marked out only by counterpredication and division it would have been left unexplained why the pre- and post-scientific classification match each other so nearly. There is no inconsistency in genera and species being determined by the method of counterpredication to which in general (30) For a contrasting view, see David Balme's Aristotle's use of division and differentiae in Philosophical Issues in Aristotle's Biology (ed.) Lennox and Gotthelf p. 80 : He argues that 'the role of Hisl. Anim. is to collect, screen and distinguish, and describe correctly the differentiae requiring explanation'. The basis for this claim is that a) Aristotle did not systematically classify sub-genera, and b) involves within some sub­ genera a wide variety of animals. Both these points seem to be correct, but only show that Aristotle had not fully used his taxonomic resources to carry through the classificatory task of discerning what groups of animals there are. They do not indicate that he was engaged in the quite different task of merely collecting animal differentiae. Balme's view, although a healthy reaction against the extreme a priori reading offered by During of the Hist. Anim., represents Aristotle as an 'out-and-out' empiricist, untrarqelled by any specific theoretical presuppositions in this work.

166

NATURAL KINDS AND NATURAL HISTORY

there corresponds no one explicable nature (e.g. the lung-possessors). If so, there is no reason to believe that pre- and post-scientific classification would match each other as well as they seem to do (in Aristotle's account). (b) It also explains why authority rests with stage I I I type considerations in determining the extension of certain concepts; for if it can be shown that there is no common nature informing a class of cases one is constrained to take them as distinct. Thus Aristotle writes in discussing pride:

it is sufficient to establish either that it has an efficient cause or is an organized structure with one teleological characterization. In both cases one may be unaware of what the relevant cause is. In the biological examples, one would establish the existence of such a structure by showing that there is an animal with a number of distinctive basic soul-functions involving locomotion, nutrition, brea­ thing or reproduction. This is to establish that a relevant differentia belongs to a wider genus in the way required to show that one is confronted with an organized structure. Counterpredication is one way to achieve this result; for it provides evidence that there is an organized nature before one. But what it is to be a common nature is what explains why, in these cases, there is counterpredication of this type. And one knows in the case of animal species and genera that such kinds exist when one knows that there is some such organized nature. Reflexion on Aristotle's method in the biological works serves to illuminate issues left unresolved and obscure in the Analytics. It is only there that one sees Aristotle spelling out in some detail answers which show what is required on his view to establish the existence of distinct genera and species. In the biological works, Aristotle does not merely follow a pre-set Analytics model of scientific investigation. Rather he is struggling to make that model at once more precise and more powerful.31

" I f we do not come to one but to two or more accounts, it is clear that what we are seeking is not a single thing but several." (Post. An. 97bl4-15)

I

I· i

I

' I

u

II

I, ri,

I'

'

167

D. CHARLES

Thus, if there is not one thing common to all proud men, there are distinct types of pride - different things in virtue of which (97b 1 7) proud men are proud. But this suggests that putative difference in internal cause (e.g. indifference to fortune: not brooking dishonour) forces difference in species or genus. But this will only be so if at the earlier stages the genera or species are marked ouL in such a way that they are dependent on (revisable in the light of) stage I I I considerations. And this will not be achieved if they are marked out by the empiricist method of counterpredication and division alone. For that method gives no ground to accept Stage I I I considerations as possessing authority concerning correct pre-scientific classification. The assump­ tion of a common nature is precisely what is needed to legitimize reposing authority in discovering the_ existence of mind with Stage I I I factors. Since in his discussion of pride Aristote places authority in taxonomic issues with Stage I I I factors, he cannot be marking out the pre-scientific divisions in ways which ignore scientific features (as he would have done had he operated only with the highly empiricist counterpredication/division method). Indeed, given Aristotle's relian'ce on Stage I I I factors, one would not expect his discussions of existence issues in these cases (stage I I) to be complete or exhaustive. It is only when the stage I I I explanatory task is finished that the taxonomic issues involved at stage I I can be finally resolved. It , is these explanatory considerations which determine (e.g.) what are the lowest levels of animal species (below which one has more/less variation within a species) and the full range of relevant species (as in the case of pride). Given this perspective, the taxonomic incompleteness of the Hisloria A nimalium was inevitable. It would not have been possible for Aristotle to have produced a complete taxonomy without considering the basic explanatory questions which he considers in detail only elsewhere (e.g. in de Partibus and de Generatione A nimalium). On this view, it is sufficient to establish that a kind exists to establish that it has a distinctive nature - even if one does not know in detail what this is. To establish that something has a distinctive nature,

(31) Earlier versions of this paper were read at the A.P.A. (Western Division) in Los Angeles in 1986 and at the N.E.H. summer school on Aristotle at Durham, New Hampshire 1988. I am especially indebted to Allan Gotthelf and Jim Lennox for many lively and interesting discussions on these issues. Jim Lennox's criticisms of an earlier version of this paper at La Rochelle, together with our daily talks on these topics in the New Hampshire, have played a major role in forcing me to rethink many of the issues in this paper. The paper has evolved considerably through this process, and contains a variety of points which have arisen out of our friendly and collaborative discussions and disagreements. I have also gained from comments of David Balme, John Cooper, David Depew, Geoffrey Lloyd, Terry Irwin, Mario Mignucci and Pierre Pellegrin. However, none of those mentioned agree with all of this paper, and some agree with almost none of it.

BIOLOGIE, LOG!QUE £r MttAPHYSlQUE

chez ARISTOTE

8eminltire CNRS-N.S.F,, 1987

Editions du CNRS, Paris, 1990.

NOTES ON DAVID CHARLES ON

i

I

HA

JAMES G. LENNOX

In David Charles' view the second book of the Posterior A nalytics describes, among other things, three distinguishable stages of inquiry; the end points of these stages are [i] having an account of what the term signifying the purported kind being investigated means [ii] having a grasp that the kind being investigated exists [iii] knowing what the kind is. Unlike certain popular contemporary accounts of meaning, Aristotle's stage [i] is neither essence referring nor existentially committed {where this means "committed to the existence of the 'kind' to which the term answers"). Stage [ii] moves the investigator to the plane of existe.ntial commitment; Stage [iii] to that of essential grasp. There is another side to these stages, according to Posterior Analytics I I . The scientific investigator also moves from knowing that certain predications hold true, to knowing why they do. That is, chapters I and 2 describe two pairs of 'things we seek'. I say 'another side' because if Posterior A nalytics I I does anything, it makes clear that investigations aimed at definitions of naturals kinds and investigations aimed at demonstrations of non-essential but necessary predications through definitional identifications are not separate processes of inquiry, but interwoven aspects of the pursuit of scientific understanding.

I

!

I

A reas of Agreement

In 'What is Natural History? ' and its close descendant 'Between Data and Demonstration . . .'1, I argued that the Hisloria Animalium is

I

(I) Forthcoming in Science and Philosophy in Classical Greece, ed. A. C. Bowen, Pittsburgh, 1990. ,

170

J , G.

LENNOX

claimed to be, and bears striking resemblances to, a report of the results of what the Poslerior A nalylics calls an inquiry into the 'thats', the explicable predicates that hold true of the various kinds of animals. I did not stress the fact that, given the close connection between the 'if it · is/what it is' and the 'that it is/why it is' pair of inquiries, it should also be concerned with establishing the 'existence of kinds', to use Charles' language. I was focused on the kinds of affirmative and negative propositions one finds in the Hisloria Animalium, and in particular on the apparent drive to establish what Allan Gotthelf and I have been calling 'widest class generalizations', commensurately universal one takes the Posterior A nalytics picture of science seriously. But if the Poslerior Analytics describes stages of knowledge acquisition which are reflected in Aristotle's biology, then these commensurately universal predications ought also to give us an initial grasp which points us toward what the subjects being investigated are as well. That is, they should count as the product of a David Charles 'Stage I I ' inquiry. A reas of Disagreemenl

The central disagreement between David Charles and me arises over the place and function of 'counter-predication' and 'division', the tools of Posterior A nalytics 1 1 . 13-18, in 'Stage I I ' inquiry. In my papers on this subject, I have argued that a report of the results of such inquiries would be arrived at through a combination of division and ' counter­ predication' - these methods would be necessary to establish the basic predications requiring explanation at Stage I I I . This claim entails, as Charles pointed out in 'Meaning, Natural Kinds and Natural History', that the basic kinds would be fixed during such an inquiry, not prior'to it, so that my argument seems to imply that these are methods which can be used to 'fix' or 'establish the existence of' kinds. There is a serious question about whether 'fixing kinds' is an actual aim of hisloria for Aristotle, or only a consequence of organizing a domain for demonstrati­ ve purposes, but for now I want to put that issue aside. While my papers insist that the 'premise selection' method of Poslerior Analytics 1 1 . 1 4 presupposes already completed divisions (and even refers to works of such which the ancient lists and other lines of evidence give us reason to believe actually existed), they also suggest that relating the terms in these divisions via the method of 'fixing problems' discussed in that chapter - establishing counter-predications - would aid us in establishing the basic predications of the science. Charles, on the other hand, wants to claim that getting a grasp on the kinds requires embarking on the study with an already established

' NOTES ON DAVID CHARLES ON

HA

171

position on what it is to have a common nature (which he thinks may derive, as G. E. R. Lloyd has often suggested, from the de Anima theory of the soul), and then having good evidence that the kind for which you have a signifying account in fact has such a nature (which you can have without knowing what that nature is). Thus on Charles' view, one begins the zoological Stage I I inquiry not only with the assumption that animals will fall into groups based on common natures, but with more robust assumptions about the constitution of such common natures. Let me then summarize the two primary differences between us: 1. Charles is focused on the HA as a document the chief aim of which is establishing the existence of natural kinds (with the discovery of co-extensive predications a secondary issue). I on the other hand am inclined to see the opposite emphasis in the HA , on co-extensive predication. 2. Charles sees the search for natural kinds as guided by a robust theory of what a common animal nature will consist in, namely an organized set of soul functions. I don't deny that many of the differentiae collected, divided and interrelated in the HA are related to soul functions, but that the treatise is organized on these grounds I seriously doubt. I see a more empiricist concept of evidence that a group of animals has a common nature at work in the HA. The more empiricist concept is this: locating counter-predications (or commensura­ tely universal predications) aids in identifying the groups that are likely to have common natures. Relalively Empiricisl Common Natures

At the risk of allowing the opposition to define the terms, I will accept Charles designation of my position as 'empiricist'. In my papers, as I mentioned, I was not focused on the issue of 'fixing kinds', but rather on 'fixing hoti-predications' as both establishing the explicanda of a domain and as a step toward establishing the explanans. But if the author of the Hisloria Animalium remains commited to the views articulated in APo. I I then he ought to be concerned both with establishing that S is P and that there are Ss. And these should be intimately connected activities. Furthermore, Charles and I agree that this is a relatively 'theoretical' stage - Aristotle tell us that at this stage, 'aware that there is a cause, we seek what this is'. [APo. 1 1 .2 90a8-9] The puzzle, of course, is Meno-like: how can we know that there is a cause short of knowing the cause? Both Charles and 'the empiricist' have answers to this puzzle - I shall supply an empiricist answer.

172

J. G. LENNOX

David Balme has been suggesting for some time now that the ultimate aim of thes.e works is to grasp the causal relations among differentiae. I would like to suggest that building up a profile of correlations among differentiae will gradually establish the existence of a kind, by providing the evidence Iha! there is a phusis common to the entities which share these differentiae. The first step in such an activity will be establishing the commensurate (i.e. co-extensive) links among differentiae. This will gradually move you toward a 'profile' of a 'bundle of counter-predicated features', which will be evidence of the existence of a kind, i.e. a group of animals with a common phusis, but not evidence of what the kind is. And this is the level of understanding that, on Charles' trichotomy, one has at the completion of Stage I I . And one manages this with only division and counter-predication as one's guide. Where might we start to defend this position? Well, PA I .2-4 has an extensive discussion of division during which there is a discussion of what it is for a collection of different animals to make up a genos. PA I.2 critiques the method of single difference divisions in 'the written divisions' for dividing in such a way that groups like Bird and Fish are broken up, part of the kind falling on one side of the division, part on the other. Aristotle insists that it is not just because a collection of animals has been given a single name that they are a genos: Now this likeness has had a name put to it, Bird, and another has But others are unnamed, such as the blooded and the bloodless: no one name has been put to either.of these. [Balme trans., 642bl4-16]

Fish.

It is a certain likeness among the members of a group that is important, not the mere act of naming. The embedding into language of names for the collection of birds and fish is praised by Aristotle in the next chapter because these names pick out groups of animals which share many differentiae at the same level of generality (kalholou diaphorai [643bl0-13]). Continuing this theme, chapter 4 begins with the question, why not go for one higher name which designates a kind embracing both the water dwellers and the flyers. 'For there are some palhi! common even to these' [644al2-16]. Yet Aristotle thinks this would be a mistake. Why? For as many of the kinds as differ by degree and the more �nd the less, these have been linked together in one kind, while as many as are analogous have been separated. I mean bird differs from bird by the more or by degree (one is long-feathered, another is short-feathered), while fish differ from bird by analogy (for what is feather in one is scale in the other). But to do this for all is not easy, since most animals are the same by analogy [644al 7-23].

When then should we speak 'of some affections in common by kinds'?

NOTES ON DAVID CHARLES ON HA

173

. . . wherever the kinds have been satisfactorily marked off by popular usage _ a?d possess both a single natur� and forms not far separated in them bird and fish and any other that 1s unnamed but like the kind embraces the forms that are in it [644b2-6].

_

He goes on to say that certain kinds have been marked off by the . likenesses of the shapes of the parts and over-all bodily configuration; the gene ment10ned here are bird, fish, 'the softies' and 'the hard shelled ones'. The last two, referred to by terms derived from adjectives which . pick out obv10us perceptual features of the things grouped are, like the first two, said to differ only by the more and the less. Nothing so far is said about certain special 'soul functions' as the relevant identities and differences for establishing kinds. Yet this discussion is focused explicilly on the philosophical question of how to make the determina­ tion of whether two different animals are one in kind. This strikes me as evidence for the empiricist position. Here the.n are two conditions for a variety of animals constituting a . kmd, cond1t10ns that will allow you to identify a kind without necessarily having grasped what is most essential in its nature. I . All members of a kind share many differentiae which are the same at a certain level of generality. 2. These general differentiae are differentiated by continuous variations of their sensible affections, i.e. by degree, more and less. . Significantly, both these conditions on kind identity are explicitly discussed at the very beginning of HA I.I 486a22-5, during its introductory discussion of sameness and difference in form, kind, and analogue. The discussion of 'variation in degree' is generally consistent with, though slightly more sophisticated than, that in PA I . N o claim is made, on the other hand, that only certain soul-function similarities are relevant. When we have seen this in a group of different animals, this may be sufficient, to use Charles' carefully chosen words, to tell us fhal we have a genos with a common nature, though not what the common nature is. But how do we recognize animals which constitute groups meeting these conditions? I s it simply intuition? Or might there be a set of methods for doing so in an organized way? On the view that David Balme has been defending for the last 25 years or so, the focus of HA is not on identifying kinds per se, but on . gettmg clear how the various differences which the animals display are related to one another. But one can now see how an identification of kinds might arise out of such an enterprise. For if a significant number of different animals differ only in degree with respect to a significant number of parts, activities and habits, so that by abstracting from those variations they are all the same, then one {probably) has a kind. This may be what counts as establishing that the kind exists. That is, what

J. G. LENNOX

' NOTES ON DAVID CHARLES ON HA

counts as evidence that there is a kind answering to a name or name-like phrase we are using is a series of commensurately universal statements (counter-predications) which either relate a series of features to the same kind as idia, establishing that it really is a kind, or to each other (thus, in a sense, making a kind). Any one of these co-extensive differentiae may be considered as good an 'explanation' of any other as 'the moon's failing to cast shadows' is of the moon's being in eclipse. That is, without more by way of causal investigation than Hisloria Animalium provides, you will have nothing more than the materials for what the Poslerior A nalylics terms holi-demonstrations. But that should be enough to make you confident that these predicative relationships have a causal explanation, even if you don't yet know what the explanation is. This constitutes a claim about what the H isloria Animalium should be doing, if it is doing something which corresponds to a pre-causal inquiry as envisaged in the Poslerior Ana/ylics. It successfully integra­ tes my earlier ideas on the role of the search for the commensurate universal in the holi inquiry, and on HA as such an inquiry, with Charles' notion that a holi inquiry (Stage I I) ought also to be concerned with establishing the existence of natural kinds. Yet there remains a major disagreement between Charles and Lennox regarding the extent to which establishing that a kind exists should involve, and in fact in HA does involve, preestablished views regarding which sorts of differentiae should be considered at the outset as relevant to a Stage I I inquiry: In arguing that Aristotle is consistently focused on a 'shortlist' of de Anima-like soul-functions, he is bucking a trend initiated by David Balme's research, which indicates a significant lack of concern in HA for the 'essence/accident' distinction in Aristotle's biology [cf. Balme 1 987a 1 1 , 19; Balme 1987c 291-312]. , Now there is a way of stating Charles' thesis which makes it thoroughly uncontroversial. For it is possible to say that virtually every feature of an animal is related to one or the other of the de Anima soul-functions. But judging by the way in which his po,sition is defended this is not the form of the thesis he wishes to defend. Rather ' it is that the central focus of HA is on the parts direclly involved with carrying out the primary soul functions. Stated in this robust way, there certainly are prima facie anomalies for the view. First, why do differences in the features which Aristotle identifies as the most common and most necessary feature of animals - perceptual organs - play virtually no role in locating Aristotle's list of large kinds? And second, why is there so much discussion of things like one kind being a different color or size from another, or bathing differently than another, or being a friend or an enemy of another, or being gregarious or a loner. The HA would be a much shorter work

were it restricted by Charles' robust theory of natural kinds. And finally, when we are told so much about the organizing principles of this work in chapters 1-7, why are we not told about this one? Further, by insisting that a central focus of HA is on fixing natural kinds, and on taxonomic considerations generally, he is bucking another trend, also initiated by Balme [1961, 1962, 1972, 1 987b] and developed significantly by Pellegrin [1982, 1985, 1986, 1987], Lennox [1987, 1990] and Gotthelf [1 988] that sees an activity preliminary to causal explanation as the central concern. On this view, taxonomy, at least as the term has been used since the eighteenth century - to imply classifications which are absolute (rather than context relative), hierarchical and exhaustive - is not an aim of the HA . To imagine that it is forces the reader to miss the actual principles by which the work is organized. As Balme repeatedly reminds us, the work is organized first and foremost by differentiae, leading to facts about gene being scattered throughout the work (note the extension of, and distances between, Bekker lines on fish in Charles' discussion). This makes the construction of a profile of one of these kinds damnably difficult, and if one imagines this was Aristotle's aim, one concludes as Charles and many before him have - Aristotle's argument is 'concealed' and a mere 'blueprint' of a taxonomy. Balme has been arguing that Aristotle is quite explicit about what he is doing, and there isn't even a blueprint of a taxonomy here, though there are lots of classifications. For the remainder of the discussion, and in the interests of stimulating further discussion, let me lay out certain of 'the appearan­ ces' about H isloria A nima/ium which any theory of its character must save. First, some negative appearances. (When I say that HA lacks this or that in what follows, the claims are governed by an 'always or for the most part' operator, as an indication of the overall character of the work: there are exceptions.) I believe at this point than many of these appearances favor the 'empiricist' reading - but at this stage of the dialectic, it is keeping these before us that is most important.

174

'

i. ,

,

;i

: 'I

i

I

175

Tilhenai la phainomena

1 . HA lacks modal operators, even though it is full of explicitly universally quantified affirmative and negative statements. The same or similar statements in the rest of the 'biological' corpus are typically modally qualified. 2. HA lacks the language of ousia, logos lifs ousias, lo Ii en einai, lo X einai, while the rest of the biology, uses this language at crucial junctures (see Gotthelf 1985a, 1 987 for a list and discussion of such passages).

' '

:I I

i

:!

I'

I.

, '

176

J . G.

LENNOX

3. Finally, it lacks, and announces that it sho uld lack, the vocabulary of causation and explanation. (The announcement is at HA I. 6 49la7-14 - it tells us the search for causes must come after the inquiry to follow). Now for some affirmations

4. Its first and foremost principle of organization is by the four main types of differentiae named in its introduction: parts, activities, sorts of lives, character traits [HA I. 1 487al 1-14]. Thus for any given animal, there may be four distinct kinds of claims about it, and within each of those kinds of differences a number of distinct claims. These will be scattered throughout the nine books. 5. Within this framework, it uses the notions of analogy, sameness in kind, the more and the less, variation in degree, sameness in form, in the way his introductory discussion of these concepts would lead one to expect, even explicitly extending it from its primary application to parts' activities and characters in Hisioria A nimalium V I I I . I .

r:

II

': !

6. As the work o f Balme [1961, rev. 1975; 1987b] and Pellegrin [1982, 1 985, 1 986, 1 987] have established, the use of the term genos holds no specific taxonomic reference; local varieties of sheep as well as the entire realm of animals can be called a genos. I suspect this means that, if we are to suppose any concern with. the identification of natural kinds in the HA, we must imagine Aristotle to be aiming at identifying scientifically significant kinds. Notice how, in discussing divisions of groups like insects into winged and non-winged (below, 8.) various popularly named groups such as bees and flies, are, to use Balme's phrase, called in only as witnesses of the difference, not as the object ·of investigation per se. Many of the references to specific kinds of fish discussed in Charles' paper function in the same way. 6. Many of the terms which refer to groups of animals all' of which have the differentia under consideration are what Posterior Analytics terms 'name-like phrases'. The Analytics explicitly seeks to extend the domain of science by allowing for definitions of groups designated by such phrases (see 12. below, and the account of definition in APo. 1 1 .10 93b29-32). 7. These phrases usually serve to 'unite' animals together for which the common user of Greek had no single name: the sheath-winged ones, the four-legged and egg-laying ones, the blooded ones, the ones with the crooked talons. The list includes �ll but three of his megista gene.

' NOTES ON DAVID CHARLES O N HA

1 77

8. There are a number of patterns of presentation of information that recur very often, thus suggesting that they reflect some underlying method. The first is exemplified by the following, which happens to be the last paragraph of HA IX. We might call this the 'division and correlation' pattern. Of the birds, there are the dustbathers, the waterbathers and those which are neither. As many as are not able to fly but are ground-dwellers dustbat4e, e.g. the hen, partridge, fancolin, lark and pheasant; and some among those with straight talons, and as many as pass their days in the rivers, swamps or seas, waterbathe; while yet others do both, that is both dustbathe and waterbathe, e.g. the pigeon and sparrow. The majority of the crooked taloned ones do neither. [HA IX 633a29-b5.]

Imbedded in this passage is the following information. a. Kind Birds b. Forms of Bird Crook-taloned, Straight-taloned, Water-dweller, Ground-dweller. c. Character Trail Manner of bathing d . Forms of Trail Dustbathing, Waterbathing, Combined. e. Correlations: I . Birds either dustbathe, waterbathe, do both or do neither. 2. As many as can't fly and are ground dwellers are dustbathers. 3. Some straight-taloned birds waterbathe. 4. As many of the [birds, slraighl-laloned birds] as pass time around water waterbathe. 5. Some birds do both. 6. The maj ority of the crooked-taloned birds do neither. Another example · of this pattern occurs in the discussion of the parts of bloodless kinds in Book IV: In addition to their other parts, the flyers among the insects have wings. But some of these are dual-winged1 as the flies are, while others are four-winged, as are the bees. None with only two wings has a rear stinger. Again, some of the winged ones have a sheath for the wings, as does the cockchafer, while some lack a sheath, as does the bee; but the flight of none of these involves the rump, and the wing has no quill, nor is it split. [532al 9-26]

Such passages as these, and there are many , seem to me both to lay out divisions, and to establish correlations between the differentiae of the divisions and between these and the animals which are themselves grouped on the basis of various different features, perhaps based on other divisions. 9. Another common pattern, at least in parts of the HA (and especially the parts on parts of blooded animals), first noted by Allan Gotthelf [cl. Gotthelf 1 988], is the use of relative clauses with hoson, as

178

I I I

·I ,

f I I

.,

I

'I

I ,

' 'i

I'

J , G.

LENNOX

the subjects of already quantified, sometimes universally quantified, main clauses, the forms being: As many as are [have] X, all/many/some/none are [have] Y. This pattern usually occurs within that described under 8., above, and thus, if one is careful, one can usually identify an implied substantive for the indefinite relative - ' As many [of the Quadrupeds] as are viviparous, all have an esophagus and a windpipe'. Through the first four books I noted 31 expressions of the form 'As many as are X are also Y ' ; 22 of the form 'As many as are X, all are Y ' ; 15 of the form 'All which (panta hosa) are X are Y; and 16 which are of various forms which don't involve universal affirmatives, i.e. 'As many as are X, few are Y/many are Y/most are Y/none are Y/some are Y, some are Z'. Interestingly, the frequency of these expressions is very high in some stretches of text and very low in others, for reasons which are not yet obvious to me. These expressions serve to express the extent of one feature relative lo animals picked out solely in terms of another feature, while leaving open the absolute extension of the correlation. JO. The next pattern that occurs over and over is the distinguishing as idia of a kind or as common both to the kind being features of to others. In the discussion of the external non-uniform and discussed parts of birds [503b29-504b l2] we are first told what they have in common with the groups already discussed [b29-34], and then many other things peculiar to birds and sub-groups of birds are noted. Explicity called idia of birds· are feathered wings [503b35], the beak [504a20], the toe arrangement and tongue of the wryneck and its ilk [504al3], and the fleshy crest of the cock [504b l l]. The chapter dealing with the external parts of fish begins, as PA I . 2-4 would lead us to expect, by saying that 'among the water-dwelling animals, the genos of the fishes is one marked off from the others, and embracing many the fish have one nature, and many forms not very forms (ideas) ' different. Fish share relatively little with the other, land-dwelling blooded animals just discussed: head, back and belly are Jl)entioned [504bl4-1 5]. By comparison with these others, it is more interesting that they have no neck, limbs, testicles, or breasts; the mention of this latter privation triggers a discussion of how widely this lack of breasts extends - is it simply the ovipara? No, even the ovo-vivipara lack them, though the dolphin, superficially like some of the sel(lchians, is pointedly noted to have them. All this goes on in a dense stretch of text from 504bl8-23. The crucial restriction of breasts is stated this way: 'Now it is not just that none of those which don't bear live generally are without breasts, nor that all those which bear live have them, but as many [of these] as bear live within themselves and do not first produce an egg.' Here Aristotle is meticulously pointing out the -

' NOTES ON DAVID CHARLES ON HA

1 79

co-extensiveness of mammae with the true vivipara. We must, I think, avoid the habit of thinking that Aristotle entered the study of animals with a classification of the animals into the true vivipara and the ovo­ vivipara. It is, I suggest, by noting correlations like this one that groups with many differentiae in common varying by the more and the less are identified. Mentioned as an idion of fish are 'gills' [504b28], differentiated among the fish by how many there are on each side, and whether they are cov ered or not, explicitly included as variation in degree by HA I 's . d1scuss10n of the concept. What, nothing else? Actually, there is suprisingly little by way of external parts that holds true for the whole group of fish. They don't all have fins [504b33-5], they don't all lay eggs externally [505b3-4], they don't all have scales [505a22-28]. So Aristotle discusses the differences in these features within the more restricted kinds which have them. 1 1 . A variation on this ' common/peculiar' pattern (which as I noted often has embedded within it the ' division/correlation' pattern) is one where Aristotle begins with a long list of what belongs to all of the most general kind under discussion, distinguishes sub-kinds of this, and notes what is peculiar to the sub-kinds. This pattern is particularly obvious in the discussion of the bloodless kinds [HA IV. 1-7]. Here, I suggest, there are many features which belong to all and only a popularly recognized kind, so that, rather than building up a profile of a kind via counter-predication, one simply establishes this group as a real kind. Interestingly, the hosa propositions are virtually non-existent in these chapters. 12. Moving now beyond 'common patterns' found in the Hisloria Animalium (without implying that even the surface has been scratched), I want to discuss a couple of passages where Aristotle notes a grouping of ammals that have not been named by one name. For it is in these passages that I think Charles' idea of the Hisloria Animalium as establishing kinds, as opposed to mine of it as establishing 'thats' to be explained, can be seen. His idea is that Aristotle has a sophisticated notion of a common nature which he takes to the enterprise. I have suggested a weaker notion than Charles', which he calls the 'empiricist' concept. Evidence of a common nature is a group possessed of many correlated differences, differentiated by no more than the more and less. He suggests, in addition, that it will be differences related to de An. potentials that will be especially a matter of focus. Now actually, this sounds as if the discovery of coextensive differentiae is at least an important necessary condition for establishing the existence of a kind, which is a much stronger role than Charles wants to give it. But let us see, at any rate, what is explicitly said in these passages. Take the opening lines on the parts of insects.

! .

I

180

J. G.

LENNOX

Let us speak in the same way about the insects. This kind exists having many forms in itself, and some which are kindred to one another have not been linked together by a common name, for example in the case of bee, hornet, wasp and all such ones, and again as many as have the wing in a sheath, for example the cockchafer, stag -beetle, blister-beetle and as many others as are such. [HA IV.7 53lb21-5J

Aristotle makes no recommendation in this passage. However, along the lines of the recommendations of PA I.3-4 for establishing a kind, it sounds as if Aristotle has his eye on two candidates for 'intermediate' groups here. The basis for constructing such groups would be an number o f forms sharing certain features not belonging to others, and variations in degree of those features. In the remainder of the discussion of the external and internal parts of insects no other features co-extensive with wing-sheaths are noted, while many features common to all insects are. In Posterior Analylics I l . 1 4 focused on the method to be used in 'grasping problems', which I take to be the process of finding the predications within a kind appropriate to explanation, Aristotle first uses an example with one of PA I's commonly (and correctly) recognized kinds, Bird. He then goes on:

I

Now at present we are speaking in terms of the traditional common names; yet it is necessary to investigate not only in these cases, but if anything else is seen to belong in common as well, selecting it we must then investigate what it follows and what follows it - e.g. having a fourth stomach and being without both rows of teeth follow having horns. Again, possessing horns follows something; for it is clear why what we mentioned will belong to them, for it will belong because they have horns.[98al3-19]

i

·:

Here, by way of an example well-known to readers of the biology, Aristotle recommends extending the method of 'counter-predication' to groups of animals that have not been named by the hoi polloi. The previous paragraph lays out the method of counter-predication by recommending positing the kind common to all (e.g. Animal), selecting what belongs to every member of the kind, and then moving to the first remaining kind (e.g. Bird) and repeating the procedure.2 Yet even where no common kind has been named, there may be recogniz able correlations among differentiae at a general level that this method of selecting from divisions and dissections may help us to identify. In so doing, we will see (perhaps) a group of animals, not commonly named, that have a number of differences in common , and forms not very different. That is, in so doing we may see groups that in all likelihood have a common nature. Perhaps this is one thing that is going on in the

i ! 1· I1

I :,

·· 1 .:

' '

11 I

' '

(2) For a more detailed discussion of this passage and the Prior Analylics 1.27-30 background, see Lennox 1987 and 1990.

1

'NOTES ON DAVID CHARLES ON

HA

181

passage which discusses the two groups of unnamed insects. (Cl. the similar passage in HA IX.40 623b5-8). Caution is called for, however: the HA never discusses the notion of a common nature in making these points. A passage in PA 1 1 !.6, which I have dlScussed elsewhere and which Charles discusses above . in note 27 certainly does, however. There he suggests that just as the name Bird applies to a certain kind on the basis of the being of a bird being ' from _ _ , , so havmg somethmg a lung belongs in the ousia of the animals that have lungs. [669b8-l 2] One has to be careful not to over-read this passage: it doesn't say, as Peck's translation does, "There is no common name which agplied to all animals that have lungs. But there ought to be, because. . . All it says 1s that what 18 common to these animals is un � ame d - Period. Still, read against the background of the Hisloria _ Ammahum, de Juvenlule and the rest of the PA on the interconnected set of organs, activities and forms of life centered around manner of cooling, it is hard not to think that Aristotle is suggesting that here is a c� mmon natur� , with v� riations of more and less across many correlated . d1fferentiae - 1.e. here 1s a true genos, as deserving of a name as Bird is.' There are two deeper problems raised by Charles' paper which I wish to finish by mentioning. I have proceeded on Charles' assumption _ that a s1gmficant part of Aristotle's energy in the earlier part of a (3) Ch� rles discusses this pass�ge in his note 27. On his reading, the passage makes . _ three essent1al po1nts which he thinks severely limit the role of counter-predication in . . . est� bhsh1n� the ex1st�nce of a genus. These are [i] that because there is no common name which applies to the kinds with bloodless lungs this fact (presumably that they have bloodless lungs) appears in t� e specification of their essence; [ii] that there is no one genus of bloodless l�ng possessors in contrast to the birds; and [iii] that having lungs appears in . the spec1fi�atio n of the ess�nce of groups marked out on other grounds. . _ I don t think th�s reading c� n be correct. First, Aristotle is explicit that there is a . s1ngle genos [genous llnos] for which bloodless lungs exist. The problem is not that these , _ d?n t c?ns�1tute a genos, but that what is common to these (presumably these animals, not kinds) is w1t�out a name. Now the reference to 'the common thing' which Jacks a name �an be read in a .numbe� of ways - the reference might go back to the lung in 669b8, and 1 "'.1ply all the animals with lungs generally; it might mean something common to animals with bloodies� and blooded lungs, in which case what is common would still be possession . of lungs, or it might refer to bloodless lungs being common to all the animals with blo�dless lungs, the group which is unnamed. The last sentence favors one of the first two options . But at �ny rate the contrast of these animals with the birds is that birds have a . name which applies to the gen?s, while what is common to this group is unnamed. And finally, the passage says nothing about lungs being in the essence of "different genera m� rked out ?n other grounds", only that 'possession of a lung is in the being of these , _ animal� which 1s somehow analogous to the way the being for Bird comes from something. On the usual reading of this passage [cf. Lennox 1987 1 17, Gotthelf 1985 313�, Balme � 972 120, 1987b84-85], Aristotle is seriously contemplating a group wider than Bird. Notice that the � em�ers of such a group would not merely share a Jung, but also many features coextensive wit� a lung: a ?eck, an esophagus, a windpipe, an epiglottis or . analogue, hfe depe�dent on air, and cooling by taking in and expelling air.

182 I

,

·

I

"

: �

'

' i ; '

Ii

I·

� I

i · I '·

J. G. LENNOX

natural inquiry will be devoted to establishing the existence of kinds with common natures. My goal has been to suggest a way in which this enterprise can be seen as complementary to what Balme, Gotthelf and I have been saying is going on in the Hisloria Animalium. But given the · use of genos in the HA, and given Aristotle's predilection for adopting 'single character' groupings of animals (e.g. those that dustbathe), there is a serious question to be raised about whether the H isloria A nimalium is interested in 'robust natural kinds' of the sort Charles is interested m. An interest in causal explanation of differentiae is quite explicitly acknowledged; an interest in definition is not. Supposing such an interest can be shown, what are the kinds that Aristotle is after? In Posterior A nalytics I I every explicit example connects definitions with demonstrations through the middle term referring to the essence of the major term in the demonstration: if the explanandum is thunder (a certain noise) in the clouds, the middle term (extinction of fire) defines thunder. How does this work with natural substances? He suggests it does in Posterior Analytics [90a9-12, 90a32-4, 93a23-4], and Metaph. Z . 1 7 104lb4 ff. wheels in the matter/form/compo­ site distinction to make it work. In thinking about animals, is Aristotle focused on grasping and defining kinds of animals, or do these gradually emerge from a primary activity focused on finding co-extensive differentiae? Finally, there is an even more troublesome nest of problems all related to questions about what the written Hisloria Animalium represents. There are at least three very different possibilities, and the role of guiding assumptions about animal natures might well differ depending on which is nearest the truth. I . The HA is a finished and organized report of an inquiry, but not explicitly pedagogical in aim. 2. The HA is primarily a teaching tool, organized in a certain way because of its place in a course of instruction for natural philosophers. 3. The HA is like a file folder, a way of storing the information from an ongoing inquiry into animals. Though it is obvious, it is easily forgotten that none of these is identical with the actual gathering of information that the HA reflects - an activity which included interviewing, observing, dissecting, reading travelogues1 works 'On Nature', other 'histories' , medical texts, and so on. And perhaps between the two, there was apparently a semi­ technical enterprise of constructing divisions. When Charles discusses HA as representative of a 'Stage I I ' inquiry (or enquiry, as he prefers), it will be worthwhile to know [a] how he conceives of the written HA and [b] whether what is true of it is true of the inquiry itself.

,NOTES ON DAVID CHARLES ON HA

183

BIBLIOGRAPHY D. M. BALME, 1961. 'Aristotle's Use of Differentia in Biology' in S. Mansion, ed., Arislote et les problemes de melhode, Louvain. (Revised for Jonathan Barnes Malcolm Schofield, Richard Sorabji, eds. 1975. Articles on Aristotle: 1. Science Lo� don 183' ' 192.) , 1962. 'G�nos �nd Eidos i:1 Aristotle's Biology', Classical Quarterly 12, 81-98. . . , 1972. Aristotle s De Parttbus Antmaltum I and De Generafione Animalium I, Oxford. , 1987a. 'The place of biology in Aristotle's philosophy', in Gotthelf and Lennox, 5-20. , 1987b. 'Aristotle's use of division a'nd differentia e', in Gotthelf and Lennox, 69-89. - , l 987c. 'Aristotle's biology was not essentialist' in Gotthelf and Lennox, 291-30 1. Allan GoTTHELF, ed. 1985. Arisfofle o n Nature and Living Things, Pittsburgh and Bristol. , 1985b. ' Notes towards a Study of Substance and Essence in Aristotle's Paris of Animals ii-iv', in Gotthelf ed., 27-54. , 1987. 'First Principles in Aristotle's Paris of Animals', in Gotthelf and Lennox eds., 167-198. , 1988. ' Historiae: Animalium et Plantarum ', in W. W. Fortenbraugh ed. Theophraslean Studies, New Brunswick, NJ. Allan GOTTHELF and James G. LENNOX, 1987. Philosophical issues in Aristotle's biology, Cambridge. James G. LENNOX, 1987. 'Divide and explain: the Posterior Analylics in practice' in ' Gotthelf and Lennox eds., 90-1 19. - , 1989. 'Between Data and Demonstration: the Analylics and the Hisloria Animalium ', in A. C. Bowen, ed., Science and Philosophy in Classical Greece ' Pittsburg, eh. 12. Pierre PELLEGRIN , 1982. La Classification des Animaux chez Arislole, Paris. , 1985. 'Aristotle: a Zoology without Species' in Gotthelf ed., 95-116. , 1986. ris�olle's �lassi{icalion of Animals (trans. Anthony Preus), Berkeley. , 1987. Logical difference and biological difference: the unity of Aristotle's thought', in Gotthelf and Lennox eds., 313-338.

�

! !

' ! '

B10LOGIE. LOGIOUI! £T M£rAPHYSIQUI!

i i :

5eminaire CNRS-N.S.F., 1987 Editions du CNRS, Paris, 1990.

:

THE EPISTEMOLOGICAL BASIS OF ARISTOTELIAN DIALECTIC ROBERT BOLTON Just as people judge the value of currencies by reference to the currency most intelligible to themselves, so is it in other areas as well. History of Animals 1.6 49la20-22 Dialectic is peirastic when it concerns matters which philosophy treats with knowledge. Metaphysics IV.2 1004b25-26

1 . CONFLICTING VIEWS OF DIALECTIC

In recent accounts of the procedures which Aristotle recommends and uses in order to reach and to justify the results presented in his works reference to his method of dialectic looms large. This is in marked contrast to the approach to Aristotle's methodology offered by leading scholars only a short time ago. Ross, for example, summed up his own account of Aristotle's view of the merits of dialectic in the following way: The discussion [of dialectic in the Topics] belongs to a bygone mode of

thought; it is one of the last efforts of that movement of the Greek spirit towards a general culture, that attempt to discuss all manner of subjects without studying their appropriate first principles, Which we know as the sophistic movement. What distinguishes Aristotle from the sophists, at any rate as they are depicted both by him and by Plato, is that his motive is to aid his hearers and readers not to win either gain or glory by a false appearance of wisdom, but to discuss questions as sensibly as they can be discussed without special knowledge. But he has himself shown a better way, the way of science; it is his own Analyiics that have made his Topics out of date. 1

(1) W. D. Ross, Aristotle, 5th ed., 1949, p. 59. Cf. F. Solmsen, Die Enlwicklung der arisl. Logik und Rhelorik, Berlin 1929, p. 26: "analytics supercedes dialectic."

186

187

R. BOLTON

THE EPISTE·MOLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

Recent writers have taken a view of dialectic quite opposed to the one summarized by Ross. Finding little or no indication that the method for finding and laying out demonstrations described in the Analylics is actually guiding Aristotle's thought in his philosophical and scientific works, scholars have turned to his descriptions of the method of dialectic, in the Topics and elsewhere, to find the key to understanding his procedures. The following summary assessments express a widely influential current estimate of Aristotle's attitude toward dialectic as a tool for not only philosophical but also scientific inqmry:

as most philosophical such as the Physics, De Anima and Metaphysics) from others in which he uses, or also uses, non-dialectical empirical methods. But even the defenders of this approach have commonly also claimed that for the discovery of first principles in science dialectic is the method which Aristotle recommends and regards as sufficient.• Generally speaking, then, the two sharply opposed views represent the dominant tendencies in recent scholarship and it will be helpful to concentrate on them for the exploration of the epistemological issues to be discussed here.

[Aristotle] nowhere suggests that any other method will lead to results which conflict with or go beyond the results achieved by the method of endoxa [i.e. by dialectic]. [Aristotle] establishes science on the basis of the opinions of 'the majority' and 'the wise'... He announces time and again that the way to the truth is through the study of 'reputable' opinions [i.e. through dialectic].2

In place of the earlier view that the method of the A nalylics supercedes and replaces the method of dialectic, the view now more dominant is that whatever other methodological procedures Aristotle may introduce none is intended in any way to supercede dialectic as the proper method of scientific or other inquiry and, in particular, as the proper method to use to discover the first principles of the sciences. By contrast, the method of searching for and setting out demonstrations which is discussed in the Analylics is commonly taken nowadays to have to do not with genuine discovery or the epistemic justification which that may involve, but only with what is required, after dialectical inquiry is completed, either to systematically display the results of inquiry, or to impart these results to learners or to deeply undersl.and these results.3 Other views of dialectic and its relation to demonstra­ tion than these two sharply opposed views have been taken. Some have wanted to distinguish those scientific works in which Aristotle's method is dialectical (including, typically, the scientific works we regard

(2) J. Barnes, "Aristotle and the Methods of Ethics," Rev. lnt. Phil., 1980, p. 495; M. F. Burnyeat, "Good Repute," London Review of Books, Nov. 6, 1986. Barnes prefers to refer to Aristotle's supreme method as "the method of endoxa," a description not used by Aristotle. But Barnes is, of course, clear that "endoxa form the subject matter of dialectic." See also M. Frede, Essays in Ancient Philosophy, 1987, p. 95 ("As with all other sciences, one arrives at [the] starting points [of metaphysics] dialectically.") For further references see R. Bolton, " Definition and Scientific Method in Aristotle's Posterior Analylics and Generation of Animals," in A. Gotthelf and J . Lennox (eds.), Philosophical Issues in Aristotle's Biology, 1987, n. 4. (3) The three views mentioned are defended, respectively, in G. E. L. Owen "Tilhenai la phainomena," 1961 (in J. Barnes et al., Articles on Aristotle I, 1975), J. Barnes "Aristotle's Theory of Demonstration" (in Articles on Aristotle I) and M. Burnyeat "Aristotle on Understanding Knowledge" (in E. Berti ed., Aristotle on Science, 1981).

2. WHAT IS DIALECTIC? A FIRST APPROACH Before considering these two views of dialectic further, it will be useful to remind ourselves in a general way of what the method of dialectic is, which has been viewed so differently by two different generations of scholars. A convenient place to begin is with the opening sentence of the Rhetoric. Rhetoric is a partner of dialectic. For both of them are concerned with the sorts of things which it is in a certain way common to everyone to know and which require no specialized knowledge. Thus everyone in a certain way takes part in both dialectic and rhetoric. For everyone on a limited basis engages in examination and in submitting to argument; and also in defending himself and accusing others. (1354al-6).

Here Aristotle points out that dialectic, like rhetoric, is a procedure for argument or reasoning on the basis nol of expert knowledge of a given science or other discipline but rather on the basis of what it is "in a certain way common to everyone to know." (1 354a2-3) The remark quoted at the head of this discussion, as well as the present passage, makes it clear that, in Aristotle's view, reasoning on such a basis is a fundamental part of the rational life of people in general. In the Rhetoric he explains how this is so. " Everyone on a limited basis engages in examination [of claims] and in submitting to argument [when under examination]." (1354a4-5) We all regularly test the claims of others in deciding whether to accept them, and submit in turn to the testing of our own claims by others, in the ordinary course of life. But this mode of reasoning which everyone engages in, and not just the learned, requires a basis for argument and a method of procedure which is suitable for use by and with everyone. This is dialectic. So just as Aristotle's interest in how the rational life of people in general is best led motivates him to describe and codify the proper technique of rhetoric, (4) See Owen 1961, p. 1 18; G. E. R. Lloyd, Magic Reason and Experience, 1979, p. 118 and T. Irwin, Aristotle's First Principles, 1988, Ch. 2 el passim.

188

189

R. BOLTON

THE EPISTEMOLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

which everyone engages in (1354a5-6), it equally motivates him to do the same for dialectic, which everyone also engages in. It is important to keep in mind what this main social setting is in which dialectic is conceived by Aristotle to have its original home, Sometimes accounts of dialectic concentrate on the undeniable fact that dialectic came to be practiced in the Academy, and elsewhere, as a quasi-sport for the purpose of mental gymnastics or training.' With this in view the dialectical method of the Topics has been described as, quite generally, involving no serious concern for truth. But in Topics 1.2 Aristotle distinguishes the gymnastic use of dialectic from two other distinct uses: the universal use, described at the beginning of the Rhetoric) in ordinary discussion or conversation in order to correct the mistakes of others and submit to the same process ourselves (IO! a27, 3034; cf. Rhel I . I !355a24-29); and a pair of uses in science - an aporetic use in raising genuine problems for both sides of an issue so that "we may more easily discern what is true and what is false" (!Ol a34-36), and an essential use in reaching the first principles of the sciences (!Ol a36 ff). If dialectic involves "no serious concern for truth"6 or,

more precisely, if it does not in any of its forms reliably provide us with better grounds for believing what is true than for believing what is false, then it is hard to see how it could be rational to use dialectic to perform the non-gymnastic functions, whether everyday or scientific, which Aristotle assigns to it. Of course, Aristotle needs to shows . us that the procedures of dialectic are ones which it is rational for us to employ for purposes other than mental gymnastics. It will be useful, in our attempt to see how he does this, to begin with a rough description of dialectical procedure. At the beginning of his official treatise on dialectic, the Topics, Aristotle makes it clear that a leading feature of the method is a special procedure for arguing for claims and for defending claims against objection. Roughly speaking, on this procedure a claim may be said to be " dialectically justified" just in case either it follows in an appropriate way from items which belong to the existing set of noted or accredited beliefs or it is consistent with (i.e., its contradictory does not follow in an appropriate way from) items in this set of beliefs. (Topics I . I 100al8-30)' The noted or accredited beliefs, or endoxa, which make up this set Aristotle limits to "the things which are accepted by everyone, or by most people; or by the wise - either by all of them, or by most, or by the most famous and distinguished." (Topics I . I . 100b21-3) The noted or accredited beliefs which belong in this collection include not only those which concern some particular subject matter such as physics or ethics but also those which have to do with what Aristotle calls logic (logike) , that is with the canons or techniques for arguing about questions of physics or ethics or anything whatever. (1.14 105b!9 ff).

(5) See G. Grote, Aristotle, 2nd ed. 1880, p. 271; H. Cherniss, Aristotle's Criticism of Plato and ihe Academy l , 1944, p. 18; J. Brunschwig, "Aristotle on Arguments Without Winners or Losers," in Wissenschaflskolleg Jahrbuch 1984/5, Berlin 1986, p. 40. The views of Grote and Cherniss are critized by Owen in Aristofle on Dialectic (1986), pp. 103 ff. Brunschwig resuscitates a version of their view. (6) Brunschwig, ibid., p. 40. Brunschw-ig argues that there are "discrepancies" and "inconsistencies" in what Aristotle says about the use of dialectic with ordinary people which "seem to show that he is uneasy" about the use of dialectic except for school gymnastics. (p. 34) He argues that Aristotle claims at 105a16-19 and 157al8-21 that "dialectical deduction (sullogismos) should be employed only with trained dialecticians," while induction is to be used with ordinary people. Since "the official subject of, the Topics is dialectical sullogismos" the Topics teaches us how to proceed "with well-trai,ned dialecticians not with ordinary people." Of course, Aristotle flatly contradicts this conclusion at the beginning of Topics 1.2. (The pragmateia of 101a26 is the one described at 100a23.} Is this inconsistent with what he says later? At l.12 105a16-19 he says at most that induction is more effective with people in general, sullogismos is more effective with trained arguers, not that one type of argument is exclusively to be used with either group. At V I I I .2 I57a18-21 he repeats the same point. (The reference here back to the earlier passage shows that mallon means more not rather.} More importantly, he repeats the point in a context which clarifies its import. The contrast between the two types of argument, inductive and syllogistic, is introduced in this context to describe two possible ways of obtaining the necessary premises for a sullogismos. (VI I I . I 155b20-22, b35156al 1) That is the topic under discussion. Later in V I Il.8 Aristotle even describes a premise of an inductive argument which yields a premise for the main syllogism as itself a "syllogistic premise," i.e. a premise concerned with the (main) syllogism. (160a35 ff) Thus, at V I I l .2 157a18 ff when Aristotle is recommending induction for use with ordinary people it is for the purpose of obtaining the necessary premises for a su/logismos. (One reason for this recommendation is given at 156a3-7, cf. 105al6-19). He recommends sullogismos for the same purpose in dealing with dialecticians. (One reason is given at 156a7- 1 1 , with 156b27-30). So Aristotle does not later contradict his claim in Topics 1.2 that one use of

the mastery of the dialectical sullogismos (100a22-4} is for discussions with people in general (101 a25-27). His later remarks only concern how best to obtain the necessary premises for such a sullogismos in discussion with different groups. (See also the references to the securing of premises at V I I I .2 157b28 ff, I58a l 4 ff.} At V l l l . 1 4 164b8 ff, as Brunschwig notes, Aristotle says that dialectic should not be used with "everyone," and gymnastic dialectic should not be used with "just anyone." Some people are too competitive in argument for this, he says. But there is no suggestion that these hyper-competitive people are the ordinary people rather than, say, mainly trained dialecticians, or some of each. (7) The successful syllogistic reasoning, from (appropriate) endoxa, for a certain conclusion, which Aristotle mentions in this passage, he typically takes to be, in a dialectical discussion, the work of the questioner. (Top. V I I l .6 160al2-I3) In this case the proposition which is "dialectically justified" is opposed to a claim which the nnswerer is set to defend. The successful submission lo argument, from endoxa, without having lo say anything inconsistent, is, in a dialectical discussion, the work of the answerer. (Top. V I II.5 159a38 ff) In this case the proposition which is consistent with the (appropriate) endoxa and thereby "dialectically justified" is a claim which the answerer is set to defend. Just what the relation between and what the value of these two modes of justification is will . be explored further below. The term 'dialectical justification' is used here in a technical sense to refer to whatever kind or kinds of justification it may be that the questioner or the answerer, or both, 'have in either of the two cases.

190

191

R. BOLTON

THE EPISTEMOLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

The Topics itself is, in fact, mainly concerned with setting out an accredited list of these canons and techniques of argument. This initial description of the method of dialectic is, intentionally, rough and in various ways imprecise and incomplete; and much will need to be done to improve it. But the description will do well enough as a starter for us to begin to see how one might categorize the method of dialectic from the epistemological point of view so as to raise some questions about its potential philosophical interest. As described so far, dialectic might well seem to embody a version of a coherence theory of justification for claims and beliefs since it takes a claim to be justified just in case it is, in an appropriate way, consistent with or implied by certain standing noted or accredited beliefs. If so, tben if it is the case, as is now widely held, that the method of dialectic is on its own totally adequate and sufficient for Aristotle for the justification of results in science, in philosophy and elsewhere, then he favors some form of coherence theory of justification in epistemology generally, a line which some people have in fact recently taken.8 Or, more cautiously, to the extent that Aristotle regards this method as sufficient to that extent it might appear that he favors a coherence theory of justification. If he does, or to the extent that he does, then it would be of interest to know what form of coherence theory he holds and why he does so; and also to know how his reasons for adopting a coherence theory may be related to those which lie behind the recent resurgence of theories of this type in contemporary epistemology. Is Aristotle's theory, like some current theories, developed in response to· the conviction that there are no incorrigible or self-certifying beliefs, due perhaps to the theory-laden character of all beliefs? Or is it developed, like other current theories, out of the conviction that appeal to what lies outside what is most fundamental in our customary scheme of accredited beliefs is somel;ww incoherent? Or does it have some other basis, different from any contemporary one?

in the Topics where he gives his clearest general description of dialectic a class of propositions - the first principles of the sciences - for which "it is not necessary to inquire why they are so; each of the first principles is worthy of belief on its own." (Topics I . I !00bl9 f) Justified belief in these propositions, it would seem, does not ultimately depend on an apprehension of their coherence with any other beliefs or propositions at all.9 However, some recent writers have argued that there is a way of understanding this doctrine so that it is compatible with a commitment on Aristotle's part to the sufficiency of dialectic for justification. Others have argued that Aristotle came to abandon this doctrine when he saw properly the value of a special kind of dialectic.10 In the latter case, it could be argued that Aristotle at least came to favor a coherence approach in epistemology11 and the question as to what motivated his change of mind would be of special interest. This is a matter which deserves further discussion. But we do not need to settle it here. Even if it is the case that this traditional problem does not in the end stand in the way of supposing that Aristotle regarded, or came to regard, the method of dialectic as sufficient for discovery and for the justification of belief, in all or in certain sciences, there are at least two other considerations which make it clear that, for him, the proper practice of scientific method clearly can and does lead to results which both conflict with and go beyond results reached by dialectic. First, Aristotle's method of discovery and of justification in natural science generally is empirical in important respects in which dialectic cannot be. Aristotle requires that scientific theories are reached and confirmed ultimately by reference to the data of perceptual observation by contrast with accredited beliefs or endoxa. (De Caelo I I l.4 303a20-23 with I I l .7 306a3-17: both quoted shortly below.) In his actual practice in his scientific works he often appeals to new perceptual data which, as he sometimes explicitly claims, contradict all standing opinion; and he uses reports of observations which come from sources (such as "some experienced fishermen") which fall outside all of the sanctioned sources of endoxa.12 Put simply, the data base

·

3. OBJECTIONS TO THE SUFFICIENCY OF DIALECTIC There is a well-recognized difficulty which needs to be mentioned at the outset which stands in the way of attributing to Aristotle a commitment to the total sufficiency of the method of dialectic in science, or in some sciences, and to the sort of coherence theory of justification that might be involved in it. Aristotle distinguishes in the very passage

(8) See, particularly, T. Irwin, "Aristotle's Methods of Ethics," in Studies in Aristotle, ed. D. O'Meara, 1981, pp. 207-8.

(9) The difficulty was discussed by J. Burnet, The Ethics of Aristotle (1900), inlr. xii. Burnet solved it by treating dialectic as a method of discovery only, not of justification. Justification was based, he thought, on intuitive awareness of the self­ evidence of the propositions in question. Most writers have solved the problem differently, (10) See M. Burnyeat "Aristotle on Understanding Knowledge'' pp. 130-132; T. Irwin, "Aristotle's Discovery of Metaphysics," Rev. Meta. 1977, and Arislotle's First Principles, 1988. (I I) Cf. Irwin, " Methods of Ethics," p. 208. (12) See, for instance, HA 513a8-15, GA 742a16-8; HA 532b18-26. Aristotle also often relies on his own new observations which clearly do not (yet) count as noted or accredited beliefs (endoxa). See, e.g. Mele. 1.6 343b8-14. For fuller discussion of these points see Bolton 1987, Sec. 2.

I '

,i

:1

192

193

R. BOLTON

THE EPISTEMOLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

which science works and reasons from includes items which do not count as endoxa. So since dialectic is restricted in its data base to endoxa13 the method of science cannot simply be dialectic. This is brought out quite explicitly in a famous passage from Genera/ion of Animals I I I . 1 0 (760b27-33) which concludes Aristotle's discussion of how bees reproduce.

For both of these reasons the definitions required by science as first principles may conflict with or go beyond those required by or available to dialectic and dialectic is not sufficient to guarantee that a definition appropriate for scientific purposes has been discovered." Again, this is a point which holds quite generally and not for some sciences as opposed to others. These two crucial features each of which separates scientific method from dialectical method (either as a method of justification or as a method of discovery) are the ones to which Aristotle standardly gives special attention throughout the corpus, in works early and late, in his summary descriptions of proper scientific procedure. The scientist's task, he repeatedly says, is to collect the full range of empirical phainomena on some subject and then look for the theoretical principles (preeminently the definitions) which serve to explain these data. Consider the following passages:

This is the way things appear to stand concerning the genesis of bees on the basis of reasoned argument and the things which are acc�pted (ta sumbainein dokounla) about bees. But the facts have not been adequately ascertained and if they ever are to be, credence must be given more to perception than to reasoned arguments, and to reasoned arguments only if what they show is in agreement with the phainomena.

Here Aristotle explicitly cites a case where the agreement of some theory with what is "accepted, " which would at least include agreement with the endoxa, is not enough. Further data he says, are required. And the primary test for the credibility of further data, and of any theorizing based on them, is not whether they become endoxa or how they fit with the endoxa but whether those data are perceptual phainomena. It cannot be claimed that this requirement is one which Aristotle invokes only for certain (empirical) sciences and not for others. In the A nalylics Aristotle explicitly invokes this requirement for "any art or science whatever." (APr. 1.30 46a l 7 ff, quoted immediately below.) Secondly, Aristotle makes it clear that proper scientific method must aim to secure explanations in a way that goes beyond what dialectic can achieve. This comes out most clearly if we consider the requirements for adequate definition prescribed by dialectic. Dialectic requires, and provides a method for reaching, definitions which are inferable from or at least consistent with the endoxa or an appropriate subset of the endoxa. (Topics 100al8-2 1 . The main lines of argument (lopoi) for dealing with definitions are given in Topics VI-V I I .) But adequate scientific definitions must satisfy a different requirement. They must be capable of explaining the phainomena. (De Anima I . l 402bl6-403a2, Prior A nalytics 1.30 46al 7-22 both quoted below.) Among these data may be items which do not count as endoxa, for reasons explained above, and the relation required between the data and the definition is stronger than mere derivability or consistency." (13) This needs certain qualifications which are, however, not important in the present context. See Bolton, 1987, n. 8. Aristotle himself usually states the point without qualification, as at Metaph. 995b23-4. (14) Of course, the endoxa with which any dialectically adequate definition must be consistent include endoxa about definitions. And it might be an endoxon principle about definitions - though it seems it was not in Aristotle's day - that definitions should serve to explain a certain range of empirical data. But even if this principle were endoxon that

Most of the principles in each science are unique to it. Therefore, it is the role of experience (empeiria) to yield the principles of each science. mean, for instance, that experience in astronomy yields the principles of astronomical science, since it was only when the phainomena were adequately grasped that the demonstrations in astronomy [and thus the principles on which genuine demonstrations must be based] were discovered. The same is true of any art or science whatever. (Prior Analytics 1 .30 46al 7-22; cf. APo. 1 . 13 78b34-79a6, I l . 19 100a6-b5.) It does not make good sense for it to turn out [as the Platonists claim] that one element alone [earth] has no part in the transformation [of the elements into each other]. Neither is it apparent on the basis of perception; rather [on this count] all [the elements] change equally into each other. 'As a result these theorists are offering accounts which concern the phainomena while their accounts are not in agreement with the phainomena. The reason for this is that they have not proceeded in the

would not permit dialectical argument to secure it that a given definition which does in fact explain such a range of data is the correct definition - even for someone who knew that the definition in question satisfied this principle. It would have to also be endoxon that the definition satisfies the requirement before it could be concluded, by dialeclical argument, that the definition is the correct one. Only endoxa can figure as premises in dialectical argument. That the principle mentioned above was not endoxon in Aristotle's day is confirmed by the fact that the Topics mentions no such principle in its list of standard procedures (topoi) for the examination of proposed definitions. It does mention the doctrine that a definition should be given by reference to what is prior and more intelligible absolutely. (Topics VI.4 141b25 ff) This doctrine is used by Aristotle in his construction of his own view that ultimate definitions should explain the perceptual data on a given subject. (Physics I . I 184a18 ff) But no such application of this doctrine appears in the Topics. The only use of the doctrine in the Topics is to argue that definition must be by genus and differentia since these are prior and more intelligible absolutely than the species. This latter point obviously reflects standing doctrine. (15) Fo:r fuller discussion of this point see Bolton 1987, esp. Sec. 5.

l :ij

I

194

R. BOLTON proper manner in the positing of their first principles because they were determined to bring everything into conformity with certain fixed ideas. For surely the first principfes which concern sensible things must conform to the sensible, those which concern eternal things to the eternal_, those which concern perishable things to the perishable: in general, principles must be conformable to their subjects. But because of their affection for their principles they have behaved like people who are set to defend their theses for the sake of argument. For holding their principles fixed as true they abide any consequence [of them] not seeing that it is necessary to judge some types of principles in the light- of their consequences, in particular of what is ultimate. The ultimate thing in the case of practical knowledge is the product, in the case of natural science it is the always authoritative perceptual phainomenon. (De Caelo l I I .7 306a3 ff.) Knowing the accidents of a thing contributes a great deal towards knowing what the thing is. For when we can give an accounting, at the phenomenal level, of the accidents - either all or most of them - we will be able to speak best about the essence. For a statement of what a thing is is the starting point of every demonstration [about that thing], so that insofar as definitions do not lead us to a knowledge of the accidents, nor even lead us easily to likely proposals about them it is clear that they have been offered dialectically and are of no value. (De Anima I . I 402b21 ff.)

The reader of such remarks and others like them (e.g., GC 1.2 31 6a513, HA 1.6 49la7 ff, GA I I I . 1 0 760b27-33) might naturally be led, as Ross and his contemporaries were, to a view of Aristotle's theory of justification in science quite at odds with the currently standard one. In collecting the appropriate range of empirical phainomena and explaining them (in accordance with the canons of the theory of demonstration) why does the scientist need to pay any attention at all to endoxa as such? What is important is to collect the proper empirical data and explain them. Whether some or all of these data are endoxa seems irrelevant; as does the matter of whether the theories one offers to explain the data cohere with the endoxa. Looking to the body of noted or accredited belief would seem to have, at most, pragmatic utility, by serving to direct attention to possibly significant data or possibly promising theories; but no capacity to guarantee discovery and no value as such for justification. What guarantees discovery and what counts for judging the correctness of theories, it would seem, is the genuine empirical status of the data explained and the genuine explanatory power of the theories. As noted above, Aristotle invokes this require­ ment not simply for the natural sciences (as in De Caelo 1 1 1 .7) but (in Prior Analytics 1 .30) for "any art or science whatever." This view of dialectic as, in the final analysis, inadequate to guarantee discovery and irrelevant for justification in science is, as we have noted, just the view taken by many scholars in the past before the current trend acquired its prominence. But despite the difficulties for the current view this older view does not at all seem to account for the

I

, ,

I

l ·L I

:

'

-----

THE EPISTEMOLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

195

role which Aristotle explicitly assigns to the use of endoxa in scientific inquiry. He thinks it is necessary to find theories which accommodate the endoxa (or at least some appropriate subset of them) as well as perceptual phenomena. In the De Caelo ( 1 1 1 .4 303a20-23) he objects to the physical theory of the Atomists because: In claiming that there are indivisible bodies they cannot avoid conflict with the mathematical sciences nor the denial both of many endoxa and many perceptual phainomena.

So he regards it as a legitimate test of the adequacy of theories in physical science that they fit not only with the perceptual phainomena but also with the endoxa. This is further confirmed by a well-known passage in the Physics (IV.4 2l la7-l l). It is necessary to try to conduct our investigation so that it is determined what place is in a manner such that the difficulties are resolved, and the things that are held to 'belong to it (ta dokounla) do turn out to belong to it, and, further, it will be clear what the cause is of the trouble and the difficulties about it. For in this way is each thing best established.

Here Aristotle claims it is necessary (though not sufficient) for the best justification to find theoretical principles which account for the "things that are held ." These, again, must at least include the endoxa, or at least some appropriate consistent subset of them. So even though he gives higher authority in justification to perceptual phenomena as such, endoxa have value in justification as well. And Aristotle says in Topics 1.2 that it is dialectic which is the proper method to use to determine how theories square with the endoxa. (!Olb l-3) He also claims in this chapter that it is "necessary", and not simply heuristically valuable, to work dialectically through the endoxa to reach adequate theoretical first principles. (!Ola36 ff) He does not restrict this requirement to some sciences as opposed to others but invokes it quite generally for "each science" . (!Ol a35) He does speak of the require­ ment as applying in "the philosophical sciences" (a34); but this does not, as some have suggested, restrict the requirement in a significant way for present purposes. Aristotle counts the scientific study of any real kind as a branch of "philosophy." (Metaphysics IV.2 1004a2-9; cf. V I . I 1026a l8 ff.) One might perhaps suppose that the task o f "dealing with the first principles," for which dialectic is essential, is a very small or special part of the scientific enterprise. Even if this were so it would, of course, be important to understand how dialectic can play a necessary role in this process. But various passages make it clear that this task is coextensive with the scientific enterprise as a whole. In the first chapter of the Physics for instance, Aristotle identifies scientific inquiry (melhodos) with the process of the acquisition of principles, and in Posterior Analylics I I . 1-2 he elaborates this point in detail. Aristotle also says that it is "the role of experience to yield the principles. . . [This]

196

.i i I1:

R. BOLTON

is true of any art or science whatever." (APr 1.30 46a 17 ff) That is, Aristotle conceives of the task of collecting and sifting the experiential data as the main element in the process of working toward the principles, for the obvious reason that the principles are found and validated as the · ultimate items that explain this body of data. (De An I . I 402b21 ff, De Caelo 1 1 1.7 306a3 ff.) Since these two processes, collecting the empiri­ cal data and explaining it, are just the ones in which Aristotle claims to be chiefly engaged quite generally in his scientific works, if dialectic is necessary for "dealing with the first principles" in every genuine science then dialectic is a necessary ingredient in the scientific enterprise quite generally. So even though the new view that dialectic is totally adequate has serious deficiencies we cannot simply go back to the old view that dialectic is irrelevant. We need to explain why Aristotle should give speciai status in scientific method quite generally to endoxa and to the method of dialectic which reasons from them, in view of the limitations of dialectic as a tool for discovery and justification in science. In particular, we need to understand in this connection why Aristotle should require the scientist to care at all about what people in general credit in' physics or biology, or other sciences. This is a question which has so far received surprisingly little attention, even from those who have regarded Aristotle's method of working through the endoxa as, for him, totally adequate for justifica­ tion in science or elsewhere. There are indications in the recent literature of several different approaches which those who view dialectic as Aristotle's primary method would take in answering the question. But it will be useful, before investigating these approaches, to consider more carefully just what Aristotle himself says about how to do dialectic. If we can see in more detail just what the basic rules of the method are we will in a better position to consider what the value may be of various recent accounts of the ultimate basis of the method.

4. H o w TO no DIALECTIC: So ME INITIAL QuESTIONS , In the opening sentence of the Topics, to which we have already alluded, Aristotle gives a general characterization of the method of dialectic. The purpose of this study is to discover a method whereby we shall be able to reason syllogistically from endoxa about any problem which may present itself and shall be able to submit to an argument ourselves without saying anything inconsistent. (Topics I . I lOOalS-21 )

In this opening passage Aristotle introduces certain paradigmatic features of dialectical method. He points out at the end of the chapter that his discussion here, and later in the Topics as well, is a discussion

' I

THE EPISTEMOLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

197

"in outline" (lupi5i), by which he means that it is not "an exact account but only . . . one which will enable us to understand each subject in some fashion." (101a21-4) This provides for us an important general cautionary note, and in particular it implies that the initial description may in some respects focus only on certain typical or paradigm cases or features of dialectic. It will be necessary to bring certain of these respects to the fore shortly. In this passage Aristotle mentions two features of dialectic. It provides us with a method for "reasoning syllogistically' (sullogidzes­ lhai) and also with a method for "submitting to an argument ourselves (auloi logon hupechonles) without saying anything inconsistent." As his language suggests, Aristotle conceives of the latter activity as carried out, in a dialectical question and answer discussion, by the answerer, the one who "submits to an argument," from endoxa, which is constructed on the basis of premises which are granted by him in response to questions, and attempts to avoid concessions which lead to a conclusion "inconsistent" with what he is set to defend. In harmony with this, Aristotle typically conceives of the former activity, "reasoning syllogistically," as successfully brought off in a dialectical discussion by the questioner. 16 This is because success on the part of the answerer in such a discussion involves, where possible, avoiding the concessions which are necessary for successful syllogistic reasoning, i.e. reasoning to a conclusion which is inconsistent with what the answerer is set to defend (or what he has granted). So in an important sense such reasoning does not even occur in a dialectical discussion unless the questioner is successful. Dialectical reasoning is reasoning "to the contradictory" of the answerer's position (Sophi·slical Refutations 1.2 165b3-4; cf. Top. V I l l . 1 1 16lb19-26. Reasoning can still occur to conclusions undamaging to the answerer, but these are not the aim of the questioner. See Top. V I I 1.6). There are then, according to the passage above, two primary requirements for the answerer in a dialectical discussion: (1) to submit to reasoning from endoxa and (2) to avoid inconsistency. But these requirements are, of course, in potential conflict. Suppose the endoxa on the topic under discussion are inconsistent among themselves. As we have noted, the endoxa include the views of the majority but also the views of the experts, and Aristotle is well aware that on topics worth discussion the views of these two groups tend to conflict, both with each other and among themselves. (Top. I . I I) If the answerer is required to submit to any valid reasoning from any endoxa then his task of avoiding (16) See above n. 6. The proper translation for sullogidzesthai is 'reason syllogistically.' For convenience below I shall often speak of syllogistic reasoning simply as reasoning. Inductive reasoning will always be described as such.

1,

199

R. BOLTON

THE EPISTEMOLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

inconsistency would seem to be impossibly difficult. By the same token, the task of the questioner would seem trivially easy. What procedure is to be followed, then, in case of conflict among the endoxa? Is each party permitted to rely on whatever consistent subset of the endoxa he may choose? Is the maximal consistent set the appropriate one for both parties? These questions, and others like them, are of special importance for determining what kind of epistemological justification for a given claim is provided (for questioner or answerer or both) by a piece of satisfactory dialectical reasoning. If any valid reasoning from endoxa is satisfacto­ ry, even on a topic where the endoxa are massively conflicting, then how can dialectical reasoning, whether in the form of a single argument or of an extended chain of connected arguments, reliably provide much at all by way of genuine justification for the acceptance or rejection of belief? Moreover, how could the simple restriction of premises to a consistent subset of the endoxa, whether the subset be maximal or chosen at will, improve matters on difficult and important topics where conflict among endoxa is quite substantial? These questions and other related ones are now basically ignored in the literature. In order to try to begin to deal with them it will be useful to simplify somewhat our description of dialectic. As we have seen, Aristotle's remarks are typically geared to reflect aspects of a live discussion in a question and answer format between two people. This, of course, reflects the history of the subject. ("Do you call the one who knows how to ask and answer questions anything except a dialectician?" Plato, Cratylus 390c). But this is one of the places where it is important to recall that Aristotle's remarks are not "exact" but only "in outline" and as such not always universally applicable to all cases but only to certain typical or paradigm cases. He points oµt himself in Topics V I I I . 1 4 (!63b3-4) that one can do dialectic by oneself, that one person can in effect play both the role of the questioner and of the answerer and produce those results which would come about if there were two parties each doing his job properly. He has in mind particularly there the use of dialectic for mental training (163a29-32; cl. 1 .2 !Ol a28-30) and the use of dialectic for the scientific purpose of seeing both sides of an issue and thereby being in a better position to discern truth (!63b9-12; cl. 1.2 !Ola35-6). But if one can properly do solo dialectic for those purposes one can also do it for the other scientific purpose of dealing with the first principles, mentioned at 1.2 !Ola36 ff. So it will be sufficient here, in our attempt to understand dialectical justification, to consider the epistemological value of results reached in a way that conforms to the joint requirements which must be met in a two-party dialectical discussion, by both questioner and answerer, whether or not there are two actual persons responsible for

those results. This is, of course, important for determining the role and the value of dialectic in Aristotle's methodological procedures in his scientific treatises since there, though he (like Plato in his uses of dialectic in the dialogues) is often engaged in debate with various actual or possible opponents, he (again like Plato) is alone responsible for the proper presentation both of his examination of the opponent and of the presentation of the opponent's position, with whatever resources for defense the latter may involve. With this in mind, we can begin to see how to deal with the questions posed above concerning the nature and the merits of dialectical justification if we compare with the opening passage of the Topics which provoked us to raise these questions, a passage towards the end of the Sophistical Refutations which is quite similar to it.

198

Our purpose was to discover a technique for reasoning about the problem before us starting from things which are as endoxon as possible, since this is the business of dialectic in the strict sense and of peirastic. But since, on account of its close affinity to sophistry, it is set up with an added feature so as to be empowered to conduct testing not merely dialectically but also from the standpoint of one who knows, for this reason we not only undertook the aforesaid business of this study, to discover how to obtain an argument on the basis of what is most endoxon, but also to discover how we can defend a position, in the course of submitting to an argument, in a similar manner. We have already given the explanation for this; for this was why Socrates used to ask questions but not to answer them, since he conceded that he did not know anything. (34 !83a37 ff)

This new description is, clearly, one which could be of use in answering the m-a in questions raised above. If dialectical reasoning, or a certain type of dialectical reasoning, must be reasoning not simply from endoxa but from what is most endoxon, or as endoxon as possible, then such reasoning cannot be based on just any consistent subset of the endoxa (even if these endoxa are agreeable to questioner and/or answerer). Nor can it simply be based on the maximal consistent subset. Rather it must be based on the endoxa which are, in some sense, most endoxon, or as endoxon as possible, at least on the subject in question. This suggests, for instance, that if the premises of an argument are endoxa, and even drawn from the maximal consistent subset of the endoxa, but they conflict with other endoxa which are more endoxon than they are then the argument based on them will not be a proper, or at least not the best, dialectical argument since it is not based on what is most endoxon among the relevant endoxa on the subject in question. By a natural extension of this one can easily conceive of a dialectical review of all of the problems and questions in a given area as serving to collect together a consistent body of information which is most endoxon on a given subject such that what is consistent with or supported in an appropriate way by reference to this information is

200

1:

201

R. BOLTON

THE EPISTEMOLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

maximally well justified dialectically. If it could be determined, then, what it is for an item, or a body, of information to be most endoxon on a given topic it could be asked what the merits of such a justification are.

this is the way to guarantee that reasoning will only be from what is "more intelligible" than the conclusion. (!59b8-9, 13-15; !60a l4-16) This shows both that dialectic for gymnastic purposes has different rules than dialectic for "testing and inquiry" (since only the latter need always reason from what is " more intelligible") and also that gymnastic dialectic need not always reason from what is "more endoxon" (since such reasoning does, and is meant to, guarantee that conclusions are derived from what is " more intelligible," and the latter simply does not always happen, or need to happen, in gymnastic dialectic). It is an obvious collorary of this that gymnastic dialectic does not need to proceed from what is "most endoxon" or "as endoxon as possible." This is clear from the fact that gymnastic dialectic need only reason from "what appears to be true." It may ignore any distinctions within this class between what is more and less endoxon and, thus, also what is most endoxon or as endoxon as possible. Reasoning from premises of the latter sort, then, is restricted to a special type or types of dialectic. What exactly is involved in this procedure? Aristotle first explicitly introduces the notion that dialectical reasoning, or a type of dialectical reasoning, should be from premises which are "as endoxon as possible," in Topics V I I I . I ! . There he ·remarks that "[dialectical] reasoning is not equally endoxon and persuasive (pithanon) on all problems." Inquiry on some subjects is more difficult than on others, "so if someone reaches a conclusion from premises which are as endoxon as possible [with respect to the problem at hand], then he has done dialectic properly." (!6l b34-38) Aristotle makes this remark at the conclusion of a discussion of the various ways in which one may criticize a dialectical argument "in itself," as opposed to criticizing the performance of those who have produced the argument. (16lbl9 ff; cl. b38 ff) Since we are here particularly interested in the merits of the arguments themselves this distinction is important for oor purposes. One of the grounds for criticism of an argument "in itself" which he emphasizes is, if the reasoning proceeds from premises which are "less endoxon than the conclusion" or "more adoxon and less credible than the conclusion."(16lb28, 30-31) This ground for criticism is, clearly, based on the requirement mentioned above which is laid out earlier in Topics V I I I .5-6 that, in dialectical "testing and inquiry," correct reasoning must proceed from premises which are "more endoxon and more intelligible" than the conclusion. (159b8-9, 13-15, 1 60a l2-16) The above remark (!6lb34 ff) is introduced as a clarifying addendum to the ground for criticism based on this requirement. (!6l b30-33) In effect, Aristotle is saying that though correct dialectical reasoning, of a certain type, proceeds from premises which are "more endoxonn than the conclusion how endoxon they are, within the class of what is ma.re endoxon, will depend on the subject at hand. Those arguments are

'

5. DIFFERENT FORMS OF DIALECTIC As an introduction to the exploration of this line of inquiry, it is useful to note that in Sophistical Refutations 34 Aristotle's de p arture from his earlier description, in Topics I . I and again in Sophistical Refutations 2, of dialectical reasoning simply as reasoning from endoxa, and his new requirement that both questioner and answerer restrict themselves to reasoning from what is most endoxon or as endoxon as possible, is coupled with a change in the description of the method from "dialectic" (or "dialectical reasoning") to "dialectic in the strict sense and peirastic." (Sophistical Refutations 34 !83a39 f) This raises the possibility that reasoning from what is most endoxon is not required in some kinds or some uses of dialectic but only in others, particularly in the special form of dialectic called peirastic. Peirastic calls for special treatment and it will be studied shortly. But independently of that study it is clear enough from other considerations that not all proper uses of dialectic do require reasoning from what is most endoxon, or as endoxon as possible. In Topics I.2, as we have seen, Aristotle distingui­ shes the use of dialectic for mental training or gymnastics from other uses. Later in Topics V I I I.3 he claims that it is not required that the answerer in a gymnastic dialectical discussion for training purposes only grant as premises things which are "more intelligible" (gn6rim6tera) than the conclusion. (! 59al0-14)17 All that is required is that the endoxa conceded "appear to be true." (!59al2)18 Whether they are more or less intelligible than the conclusion does not matter. In Topics V I I I.5, however, Aristotle says that in dialectic used for "testing and inquiry" (! 59a33) it is required that the answerer concede only what is both "more endoxon" and " more intelligible" than the conclusion; and he adds that the point of conceding only what is more endoxo n is that (17} Aristotle contrasts here procedures for training with those for "learning." Whether the latter is or is not understood here as a species of dialectical procedure requires discussion. (See below n. 21.) This does not matter for present purposes; all that is relevant here is what is not required in gymnastic dialectic. The difficult term gn6rim6teron is usually translated here 'more intelligible;' the intelligible is not simply what can be understood but what somehow makes sense. (18) This does not mean that they must be accepted as true by the ans,ver and/or the questioner. In Top. VI I l . 1 1 Aristotle points out that in dialectical discussion for "training" one sometimes is required to argue from false premises. But still these premises should be things that "appear so." (lo phainomenon, 16l b4)

i

!

I

I

1,

'I

I

i':

202

R. BOLTON

most commendable which proceed from premises which are maximally endoxon and (thus) maximally persuasive. But sometimes the best dialectical argument, or arguments, that can be p roduced for some _ conclusion, or on some subject, will be ones for which some person is · to be commended for having done the best that can be done on that subject in the sense that there are no more endoxon premises on that subject from which the conclusion follows, and perhaps even . no more endoxon premises on that subject which conflict with the premises used; but still the argument may be less good than some other argument on some other subject which has premises which are, absolutely speakmg, more endoxon. The interest of this, for present purposes, is twofold. First, it supports the suggestion that Aristotle's interest, in Sophistical Ref�fa­ tions 34 in dialectical argument from what is "as endoxon as possible 1s an inte�est in the special type(s) of dialectic (viz. non-gymnastic dialectic) in which the premises must be "more endoxon" than the conclusion since that is his interest in the passage where he introduces this notio�. Secondly, and more importantly, it shows that Aristotle is prepared to apply two different standards for judging the merits of such dialectical arguments - a personal standard and an impersonal standard. By the personal standard an argument may be commendable because it is an argument from what is more endoxon than the conclusion, and as much more so as possible given the subject under discussion. Nevertheless, it may be, judged simply as a dialectical argument by the impersonal standard, quite weak because its premises are, in an absolute sense, not maximally or highly endoxon. It is in this context that Aristotle makes his often misunderstood remark that: It is unjust to find fault with those who reach a true co�clusion from . _ false premises; for what is false must always be syllo�1st1cally 1nferr�d . from . false premises and even what is true can sometimes be syllog1st1cally inferred from false premises. (162a8- l l ; cf. 161a24 ff, 126b27)

This is not, in Aristotle's view, an unjust criticism of a dialectical argument in its own right. Dialectical arguments can be criticized or defused (luein) for having false premises (V I I I . I I 161b2 1 ; 10 160b5, 2339); and a dialectical argument with false premises is a "bad argument." (V I I I . I I 161b7 with 161a24 ff) It is, sometimes, an unjust criticism of the performance of some person who has argued from premises which are, though false, as endoxon as possible on some difficult and problematic topic. (VI I I . I I 161 b6-8 with a24 ff) But the merits of the performance do not mitigate the demerits of the argument. (b7) So this remark cannot be used, as it often has been, to show that dialectic is an unsuitable procedure to use in the search for truth. To understand how dialectic can be so used, however, our first concern must be to

THE EPISTEMOLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

203

understand what it is for a dialectical argument to be, by the impersonal standard, as commendable as possible. That is, we want to know what it would be for a dialectical argument to be based on premises which are "most endoxon" in an absolute sense. As a prelude to this, it will be useful to consider, in more detail than is usual, how a dialectical discussion w� rks to achieve the result that reasoning proceeds from what is most endoxon, or as endoxon as possible, at least on the subject in question. 6. GIVE AND TAKE IN DIALECTICAL INQUIRY Sometimes it seems to be supposed that a dialectical discussion is a very formal affair where there is little or no room for give and take. The questioner asks whether the answerer will concede certain premises and the answerer, unless perhaps he has a question about the meaning of a premise, must say simply "yes" or "no."19 Questioning proceeds until enough concessions are extracted to produce an argument which is then evaluated in isolation from other arguments and claims. Aristotle perhaps gives this impression in Topics V I I I.7, but it is clear from the surrounding chapters that a dialectical discussion typically involves much more give and take than this. From the consideration of how this give and take normally goes, we can see how a dialectical discussion draws on, or comes to draw on, premises which are "most endoxon," on the subject in question. To begin with, each of the premises necessary for the questioner's proof of his thesis is standardly introduced not simply on its own for acceptance or rejection but as the conclusion of an argument employing induction or analogy or reasoning. (VI I I . I , 1 56a3-l l , V I I I.8 160a389) The main exception to this is in the case of universal premises which "appear thoroughly to be so" (lian prophaneis, V I I I . I 155b37; cf. V I I I .2 158a3-6.)20 If a premise is apparently so beyond any doubt or question then there can be no worries that an answerer will fail to concede it. In other circumstances the situation is different. There an answerer may fail to concede a premise. But if he does so he is not simply expected to say "no." If a general premise is reached by inductive argument based on instances which are "apparently correct" (phainomena), then if it is rejected by the answerer an objection (enstasis) or counter instance must be provided by him. (VI I I .8 160b3-5; cf. VI I I .2 157a34 ff; b313) However, if the counter instance is a good one, or at least appears to (19) See Brnnschwig (1984/5) p. 33. (20) Cf. lian periphanes, 158al . Here \ve are, no doubt, dealing with what is most endoxon in an absolute sense. For further discussion of this see below.

I1i1 1

205

R. BOLTON

THE EPISTEM.OLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

be (160b2), the questioner is not required to simply abandon his general premise. Rather he should qualify it so as to exclude the cases covered by the objection (VI Il.2 157b9 ff); unless, possibly, he is able to produce other premises to defeat the objection or support his preplises. (VII I . I 156a38 ff; VII 1 . 3 159a4 ff) I n either event, nothing prevents the answerer from then introducing new objections, which can be responded to in turn. Sometimes the questioner will aid the answerer, in order to gain his confidence, by bringing objections against some of his own premises. (VII I . I 156b 18-20) In addition to offering counter-examples to defeat inductive arguments for a questioner's premises, an answerer can introduce independent arguments (anlepicheirein) against these premises (even after they have been granted and a damaging .conclusion has been drawn from them), in order to undermine (luein) the conclusion based on them. (VI I I .8 160b5- 10 with V I I I .9-10) Moreover, even if this sort of lusis or undermining cannot be provided, an answerer may straightforwardly argue against a conclusion in its own right, where it conflicts appropriately with accepted opinions. (VI I l .8 160b6-IO) When he produces such arguments, against premise or conclusion, the answerer is, of course, in effect operating as a reasoner, i.e., as a questioner. Given this, there is no reason why his argument, in objection either to premise or conclusion, cannot in turn be objected to by the original questioner who at this point is operating as an answerer.21 What rules determine when an objection (enstasis) or an under­ mining (lusis) or an attempt at counter-argument (antepicheirein) is a good one in such a give and take discussion for the purpose of inquiry? Clearly enough, the main rule will be the same one which governs all such dialectical reasoning, namely that the argument, if relevant to the matter at hand, must proceed from what is "more endoxon" than its conclusion. (V I I I .5-6, see the passages quoted in the next paragraph below; 1 1 161b30-I) This means that if a counter­ example is offered to a universal premise it must be more endoxon than

that premise; if the counter-example is, in turn, objected to it must be by reference to what is more endoxon than it. If a premise or a conclusion is subjected to undermining or counter-argument it must likewise be on the basis of premises which are more endoxon than it, and any counter response to this must be based on what is more endoxon than these premises. Thus conceived a single dialectical inquiry not only involves an attempt to establish some claim by reference to endoxical premises which are (I) more endoxon than that claim, but which are also (2) such that there is nothing more endoxon than these premises with which they conflict. Such endoxon premises will be relatively more endoxon than certain other things but not less endoxon (or more adoxon) than anything incompatible with them. In that respect these premises can be said to be dialectically undefeatable. As such they can be said to belong to what is as endoxon as possible or what is most endoxon, not simply relative to some particular conclusion drawn from them but also relative to other endoxa on the topic which they concern, including any endoxical information in any way relevant to that topic. Given the discussion in Topics VII I, there seems little doubt that Aristotle's interest in Sophistical Refutations 34 in a method for arguing from what is "as endoxon as possible" at least includes an interest in the method for achieving arguments from such premises which he spells out earlier. Nevertheless, as we have seen, Aristotle also supposes that among the dialectically undefeatable premises on different topics some will be in another respect more endoxon than others so that the most endoxon of all would be a subset of the undefeatable propositions. To see whether Aristotle has any special interest in these propositions in Sophistical Refutations 34, and to see further what the epistemological status of this special class of propositions would be, it is necessary now to turn to the crucial question of what it is for one proposition to be more endoxon than another.

(21) This elaborate procedure where premises are not simply accepted or rejected but subject to dialectical objection and scrutiny is one which Aristotle says is inappropriate when one is doing dialectic for gymnastic or training purposes. Rather, he says, it is to be used by "one who is trying to learn." (V I I l .3 159a4-14 at all) It has been supposed that in saying the latter Aristotle is indicating that this procedure is not to be used in dialectic but only in (non-dialectical) discussion for learning. (See E. Poste, Aristotle on Fallacies, 1866, p. 102) But this is impossible. In the passages referred to above, Aristotle clearly has in view the use of this elaborate procedure in dialectic itself. This is one clear indication that Aristotle thinks that a certain kind of dialectical discussion can be used for learning. It is worth noting, however, that this learning procedure is not the one in view at SE. 2 165bl-3, which Aristotle distinguishes from dialectic, since there the answerer who learns never objects to the questioner/teacher's premises but always accepts them (as Poste sees, p. 102).

The main evidence for answering this question is found in Topics V I I I .5-6. There Aristotle lays down the principles which govern dialectic when used for "testing and inquiry." (159a32 ff) His main general principle there is this:

204

I'

7. REASONING FROM WHAT 1s MosT ENnoxoN

The one who reasons correctly [as a result of questioning] establishes his set thesis on the basis of things which are more endoxon and more intelligible [than the thesis itself]. {159b8-9)

Further on he adds, in the same vein: The things which are granted [by the answerer] must all be ... more

endoxon than the conclusion, if the less intelligible is to be reached through

the more intelligible. {159bl3-15)

I I

Ii I1 1

I

'

206

R. BOLTON

And: The questioner shall complete his reasoning :w ith all t�ose things being conceded to him which are more endoxon than his conclus1on. Those who try to reason from things which are more adoxon (discredite�) than the · conclusion cle.arly do not reason properly. Therefore, such things should not be conceded to questioners. (160a12-16)

:!

These passages all confirm that one endoxon can be more endoxon than another and they suggest what is later confirmed in V l l l. 1 1 , that the more endoxon some claim is the more weight it has in at least one type of dialectic. So the preferred claim, in case of conflict, in dialectic will always be the more endoxon claim; and the most preferred are, thus, the most endoxa. Which are these? Our passages here enable us now to answer this question. As we have already noted, Aristotle couples here what is more endoxon with what is more intelligible (gn6rim6teron); and he claims that "if the less intelligible is to be reached through the more intelligible" then we must proceed to the less endoxon through the more endoxon. This shows that the more endoxon beliefs will always be more intelligible than the less endoxon beliefs. But more intelligible in what sense? That depends, for one thing, on the type of dialectical inquiry we are pursuing. If the discussion is ad hominem and concerns the position of some particular individual who is to be examined from his own point of view, then we are dealing, as dialectic always is, with what is endoxon, but only endoxon to that individual, and our arguments must be from what is more endoxon and hence more intelligible lo ihal individual: If the position in question is endoxon or adoxon not simply but to the answerer then what ought to be conceded or not is judged by reference to what is held or not held by him. Or if the answerer is set to defend the opinion of someone else, it is clear that each point should be conceded,or refused with a view to that person's own mind. (159b25-29) ·. But dialectical argument need not be ad hominem in either of these ways. More often it is based not on the endoxa according to some individual but on the endoxa without qualification - "the things which are accepted by everyone, or by most people, or by the wise'." Here what is more endoxon and thus more intelligible must be what is so not in relation to any particular individual but somehow generally. Earlier in the Topics (as in various other places) Aristotle describes two general ways in which one thing may be more intelligible than another, in neither of which is a relationship to some special individual {or group) in question. One thing may be more intelligible than another "simply" or "to us." Thus, simply, the prior is more intelligible than the posterior; a point, for example, than a line, a line than a plane, and a plane than a solid; just as a uniL is more intelligible than a number - since it is prior to and the basis of every number. Similarly, a letter is more intelligible than a

THE EPISTEMOLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

207

syllable. However, lo us it sometimes turns out in reverse. For a solid falls most of all under our perception and a plane more than a line and a line more than a point. For people in general come to know such things earlier, since they can be understood by the ordinary sort of intelligence, the others only by an intelligence which is exact and uncommon. (Topics VI.4 14lb5-14) Does proper dialectical argument proceed from what is more intelligible {or what is most intelligible) simply or lo us? Pretty clearly it must be the latter. As we have seen, dialectic is described in the opening sentence of the Rheloric as like rhetoric in drawing on things "which it is in a certain way common to everyone to know." {!354a 2-3) Things which are more intelligible simply, as described above, do not fit such a description. They are intelligible only to an "exact and uncommon intelligence." The things which are most intelligible simply include as paradigms the first principles of the sciences as such. {Cl. Physics I . I 184al6-23) But these cannot be as such among the most authoritative premises for dialectic. For these need not even be endoxa, much less most endoxa. They need not be in every case (or any case) known or accepted by appropriate parties. But endoxa must be things which are accepted. The implications of this last fact are most clearly brought out in a passage in Posterior Analyiics 1.19: It is clear that those who are reasoning from the point of view of standing opinion (kaia doxan}, that is only dialectically, should only consider this, whether their argument proceeds from the most endoxa premises possible. Thus, although a term is not in truth the middle term between A and B, if it is taken (dokei) to be then one who reasons through it has reasoned dialectically. But with a view to truth it is necessary to proceed from what actually holds. (81b18) As Aristotle makes clear here, proper dialectical reasoning, as reasoning from what is most endoxon, need not proceed (in a case where a deductive syllogistic argument is being used), though what is "in truth the middle term." Rather it need only proceed through what is accepted as the middle term.22 This is to say, however, that the explanatory relations which actually hold among things need not be reflected in a proper dialectical argument, since an argument through a true middle term is just one which gets these actual explanatory relations right. (Posterior Analylics 1 1 .2 90a6-7) This shows that it cannot be a requirement of dialectic that its arguments proceed from premises which are more or most intelligible simply. For arguments which follow such an arrangement cannot fail to reflect the natural order

(22) For a term lo be accepted as a middle-term is presumably for it lo figure in two accepted proposition in the appropriate way and for the syllogistic conclusion from the two propositions lo be accepted as holding by virtue of the truth of the two propositions. See APr. I.I 24b18-22.

1�i·:.iI Iii

i,\:

ii

1: i

208

R. BOLTON

of explanation, and so to exhibit true middles, when we are dealing with matters where there is such an order. (Physics I . I 184a!0-21) Given this when Aristotle characterizes dialectical reasoning as proceeding fro r:i what is more endoxon and thus more intelligible he must have in , mind what is more endoxon and more intelligible "to us." This is confirmed by a passage in Sophistical Refutations 33 where Aristotle directly equates what is "held most of all" (malista dokounta) with what is "endoxon most of all. " (182b37-183a4)23 8. RANK ORDERING AMONG ENDOXA Generally speaking then, those endoxa which are most endoxa and which, thus, have the most weight or authority are those which have the greatest level of actual (explicit or implicit)" acceptance. In c ase of . conflict, this is the rule to be used to describe which endoxa to retam and which to reject. Given this, it seems likely that the order in which the different types of endoxa are introduced in Topics I . I 100b21-3 is not accidental. The first type introduced is "the things which are accepted by everyone." These are the things which have, in dialectic, the greatest weight. This is confirmed by the fact that Aristotle tre �ts what is apparent to everyone as practically unchallengeable. (Topics V I I I . I 155b37 with V I I I.2 158al)25 What everyone holds, by way of affirmation or denial, is thus most endoxon. Does this put what everyone holds absolutely beyond challenge? Aristotle comes close to saying this but he does not ever quite say it; and clearly he should not say it. Things which everyone holds could be themselves inconsistent. Here a resolution could be reached if among the things which are acceptable to everyone some are (23) Elsewhere in the Topics Aristotle can simply equate what is more intelligible with what "we know better." "A proprium of a thing must be given by reference to what is more intel!igible than the thing. For instance, someone who claims that "being most like the soul" is a proprium of fire refers to the sciul, and this is less intelligible than fire, because we know better what fire is than what the soul is." (V.2 129b9-12) (24) On implicit beliefs as endoxa see Barnes, "Methods of Ethics," p. 501. (25) Certain passages in the Ethics, assuming the method there is diale-ctical, further confirm this: "Some object that what everything aims at is not thereby good; but there is nothing at all in what they say. For what is accepted by everyone that, we claim, is so; and one whose purpose is to undermine this conviction will hardly have anything more credible to say.'' (Nicomachean Ethics X.2 1 172b35-73a2) " Is it then practical wisdom which resists [and is overcome in incontiOence], since it is the strongest state? That is absurd since the same person will be at once practically wise and incontinent on this view; and not a single person would venture to say that someone who is practically wise is the sort of person who would do the worst things willingly.'' (Nicomachean Ethics V I I .2 1 146a4-7)

THE EPISTEMOLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

209

more intelligible to everyone than others and thus more firmly held. These could count as the more endoxon (the more noted or accredited) not in extent of acceptance only but in strength of acceptance as well. In any event, not all endoxa are accepted by everyone. The next class which Aristotle lists is "the things which are accepted by most people." These come next, quite clearly, in the hierarchy of what is "more endoxon and more intelligible" to us. Here too there can be conflicts needing resolution among beliefs which are equally widely accepted and, thus, here too strength of acceptance as well as extent of acceptance might need to be appealed to. Next in Aristotle's list of endoxa come "the things accepted by the wise." (sophoi) Who does he mean by "the wise" here? In some contexts he uses this term to refer to those who actually have special know-how or scientific knowledge and understanding. (Nicomachean Ethics Vl.7 l 141a9-20) But he does not use the term in this objective sense here. He substitutes for 'wise' in his discussion 'someone famous as a philosopher' (Topics I . I I 104bl 9-20; cf. with 104a8-12 and b31-4) ; and it is clear that the views of those designated by the latter description are not necessarily for him instances of actual knowledge. As examples he mentions Antisthenes' view that it is impossible to contradict anyone and Melissus' view that being is one (b20-2). So "the wise" here are those who have the recognized standing of experts or savants but not necessarily those who are such. (We continuei of course, to use terms such as 'expert' in this way in referring, for example, to "what the experts say.") This makes it clear how the views of "the wise" can count in their own right as noted or esteemed views (endoxa) in a way in which the views of individuals with actual wisdom but who are totally unknown could not. The views of recognized experts are those which acquire standing for us in the body of recognized "wisdom." This is also important for understanding how the views of "the wise" can be less authoritative in dialectic than the views of everyone or of most people. That they are less authoritative is further confirmed by Aristotle's discussion in Topics l. 10-1 1 . Aristotle introduces there the notion of a dialectical premise (protasis), as a proposition which someone could be prima facie expected to grant when the question of whether to accept it is raised in dialectical discussion. He describes it as follows: A dialectical premise (prolasis) is a proposition, introduced in the form of a question, which is endoxos (esteemed) by everyone or by most people; or by the wise - either all of them, or most, or the most famous, assuming it is not paradoxical;26 for anyone would grant what is accepted by the wise if it is not opposed to general opinion. (1.10 104a8-13) (26) At 104b2� enanlion tais doxais is substituted for paradoxos (cf. b l 9). Trans­ lators have sometimes used 'contrary to general opinion' or the like to translate

210

i!'

21 1

R. BOLTON

THE EPISTEMOLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

Here Aristotle makes it clear that in dialectic it is not to be expected that an opinion of recognized experts will be granted when it conflicts with dominant maj ority opinion. (Cf. Polilics I l l. I I 1282al 5 ff) Thus, such an opinion does not count as a premise for a · dialectical argument. This fits with what we would expect given that dialectical argument has to proceed from what is more credible to and intelligible to us. We readily credit the recognized views of experts on matters where we have no independent opinion of our own. There the views of recognized experts are the most widely recognized and credited, and thus the most intelligible to us, of available views. Where experts have views which conflict with our own preponderant view, however, their views are not the most widely credited and intelligible to us of those available. Thus they have less authority in dialectic where this is the standard in play. Of course, the view of an expert still counts by definition as an endoxon (100b21-3), even if it is contrary to the majority view. How then does it enter in as an item with some weight in dialectical discussion if it cannot serve directly as a premise? Won't an answerer always be entitled to reject it out of hand since it conflicts with the more endoxon majority view? Aristotle's practice does not fit with such an approach; and his theory in the Topics shows how the views of experts even when they conflict with and are on their own less endoxon and less intelligible than majority views, can be introduced and even successfully defended in dialectic. An expert view which conflicts with majority views is called by Aristotle a lhesis. A thesis is a view which is paradoxical, held by someone famous as a philosopher... Or a thesis may be a view opposed to received opinions for which we have a reasonable argument; for instance the view maintained by the sophists that not everything that exists has either come to be or is eternal. For the cultured person who is literate exists though he has not come to be and is not eternal... This view, even if it does not seem [on its own] true to someone, might well seem so because of the supporting argument. (104bl 9-28)

offer reasons and explanations for what they claim. (Melaph. 1 . 1 -2 98la.24 ff, 982bl l ff ; SE I 164a l l -28, 165a!9-28) Certain passages in the Ethics confirm this connection. It is perhaps rather pointless to examine all views; it is sufficient to examine those which are most prevalent or those which are . taken to be supported by some reasonable argument. (1 .4 1095a28) Of the two types of views which Aristotle counts as worth considering here the minority views held by experts would belong to the second; and later at 1098b27-8 he redescribes the two types of views which he has proceeded to consider as those either of many people or of the renowned. From this it is easy to see how special expert views contrary to more widely accepted or majority views can be taken seriously and even established in dialectic without contravening the general rule that what is credited by and more intelligible to most people takes precedence over the views of the wise. If the expert views can be seen to follow from universal or majority views then they acquire the authority of those views and so can pose a challenge to other (less endoxon) majority views. Where there is no dominant majority view on some subject, however, the views of the wise are, as we have seen, the most authoritative just on their own. It is of interest that Aristotle gives no special weight at this juncture to the opinions of a bare majority. The opinions of most people (hoi pleisloi) or of the general community (hoi polloi) outweigh the contrary views of experts (104a8-13), but the opinions of a bare or small majority do not. This is intelligible in the light of Aristotle's general scheme. Dominant community views will be widely credited by us, but where there is a close division among community views, Aristotle supposes we are likely to credit the "wisdom" of the experts rather than the views of a small general majority. The views of the wise can also conflict with each other. In introducing the views of the wise in his list of the types of endoxa, Aristotle mentions first the views of all the wise, then those of most, then those of the most famous and distinguished. Is this also a ranking in order of greater to lesser authority? Are we to understand Aristotle as saying that if the experts do not all agree then we should rely on the view of most of the experts, and if there is no substantial majority view only then should we prefer the views of the most famous and distinguished to those of other experts? Though Aristotle says nothing explicit on the point, this would make sense given his general approach. He says, as we have seen, that we follow the practice of relying on established expert views where we have developed no relevant common views of our own. But if we have developed no views on our own and we are thus in the position of having come to rely on

This passage indicates how a view which is, on its own, contrary to general opinion can acquire standing in dialectical discussion. It can be introduced by use of an argument which shows that it follows from what we would generally accept. Aristotle does not explicitly apply this point to the views of philosophers here, but the context suggests that he has this in mind; and he typically thinks of philosophers, or equivalently "the wise" (sophoi, 104b33), as those who paradoxos. This has the strong authority of 104a8-12 behind it, and that passage at least indicates that this describes a clear case of a paradoxos view for Aristotle and the case which is relevant for the characterization of a dialectical protasis. But 104b31-34 suggests (with b19-28) that a paradoxos view of an expert may also be one which conflicts with the received view of the (other) experts, where there is no conflict with general opinion.

212

THE EPISTEM.OLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

recognized expert views as the ones we most credit it makes sense that the view on a given topic held by most experts would be the one which · has achieved the greatest renown and credit by us. If, however, there is no dominant view of the experts (again a bare majority counts for nothing), then it makes sense that we should go with "the most famous and distinguished." That is, we · should go with the experts whose views have received the greatest recognition from us. (The word translated 'distinguished' is endoxoi, 100b23.) This again seems the best we can do if we are relying on what is most well-accredited by us. However, in all these cases as in the earlier ones, it would, of course, be possible to over-ride the prima facie order of authority by appeal to a supporting argument for some claim where the premises of that argument belong to a more authoritative category than the claim. If this account is correct then the initial order in which Aristotle lists the types of endoxa is not accidental. He moves in his list in order from the (prima facie) more to the (prima facie) less authoritative types of endoxa.

Topics, that dialectic does play its necessary role in scientific inquiry mentioned in Topics I . (1.2 101a36 ff) Writers have standardly suppo­ sed that Aristotle simply does not say anything further about this in the Topics. Some have even suggested that this is an indication that the method of the Topics itself is not one which Aristotle in fact conceives of as appropriate for use for the scientific purposes mentioned, but only for use in school gymnastics.27 It is, however, a mistake to think that Aristotle says nothing further about this matter. In what we call the Topics he does not elaborate on his claim that it is necessary to use dialectic to reach the first principles of the sciences. (He comments, very briefly, in Topics V I I I . 14 163b9 ff, only on the other scientific use mentioned in 1 .2 !Ol a34-36.) But in what we call the Sophistical Refulalions, which is clearly a part of the Topics," he is particularly interested to describe a special form of dialectic which he calls peirastic (peirastike) or the art of testing. Peirastic is introduced as a method for reasoning from "what is necessarily known by anyone who professes to have episteme" (i.e., technical or scientific knowledge, SE 2 165b5-6). Later it is described as a procedure which, though it does not require episteme and can be used even by the unlearned, nevertheless can be used to test the claims of all those who profess episteme in whatever subject (SE 1 1 I 72al 7-32). In Topics 1.2 Aristotle j ustifies his claim that the use of dialectic is necessary to reach scientific first principles on the specific ground that dialectic is "capable of examination" (exetaslike, 101b3). The two words peira­ slike and exelaslike, here translated 'testing' and 'examination,' are derived from roots which overlap in meaning in Greek and, though peiraslike is a technical term, various cognate forms of the two words are used interchangeably by Aristotle (as they are by Plato) in his gene­ ral descriptions of dialectical procedure. (Rhel. 1 . 1 1354a5, Top. V I I I.5 159a25, 33; cf. Plato, Apo/. 22e-23c, Prof. 348a) If Aristotle had used peiraslike, or some cognate word, instead of exelaslike, in Topics 1 .2, then the material in the Sophistical Refutations would long ago have been thoroughly studied to understand Aristotle's famous doctrine in Topics 1 .2. But it would appear that little attention has been paid to the apparent connection between the Topics passage and the later material up to now.

9. DIALECTIC AND PEIRASTIC IN SCIENTIFIC INQUIRY

"

I

213

R. BOLTON

We are in a position now to raise pointedly the main question raised earlier: Why does Aristotle think it is necessary in scientific inquiry to use dialectic? Consider the recommendation he gives to the inquiring dialectician. As we can see now, in 'Offect it comes to this: Base what you say, if you can, on what everyone accepts; failing that, appeal to what the dominant majority think. If there is no dominant majority view appeal to the view of the experts. If the experts do not agree, appeal to the dominant majority of the experts. If there is no dominant majority view among experts appeal to the view of the most famous and distinguished experts. This sounds like appropriate advice for the adman trying to maximize sales or the politician trying to maximize support for his policies, but hardly for a scientist. That is to say it seems to embody a procedure for j ustification which is designed to be maximally effective in producing conviction when used with people in general. This is, of course, just what is to be expected since, as we have already noted, Aristotle thinks of dialectic as like rhetoric in focussing its appeal on what everyone is maximally ready to understand and accept. But the aims of science are different. The scientist's results can, Aristotle says, only be grasped by "an intelligence which is exact and uncommon." (Topics VI.4 141bl3-14) So why should science have any interest at all in procedures for justification best designed to produce maximal understanding and conviction in people in general? In attempting to answer this question it is important to try to determine just how it is, on Aristotle's own theory of dialectic in the

·

10. S OME MAIN FEATURES OF PEIRASTIC What is peirastic and how does it serve to examine or test claims to scientific knowledge? Aristotle describes peirastic as "a part of (27) See Brunschwig (1984/5) p. 34. Cf. Ross, Ar;stot/e, p. 57. (28) The evidence for this is conveniently assembled by Brunschwig, Arislole: Topiques, I (1967) p. xix.

214

I

.I "

(I j' I

215

R. BOLTON

THE EPISTEMOLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

dialectic" and "a type of dialectic." (SE 8 169b25 ; 1 1 I 71 b4-5 ; cl. 9 I 70b8-l l ) Thus, like all dialectical arguments, peirastic arguments employ endoxa as premises. (SE 2 165b3-4) They must, however, also satisfy certain stronger requirements. Aristotle initially defines peira-· stic arguments as those whose premises are "things which are held by the answerer and which are necessarily known by anyone who professes to have episti!mi!." (SE 2 165b4-6) Some have claimed that this explicitly distinguishes peirastic arguments from dialectical arguments and contradicts Aristotle's later claim mentioned above, that pefrastic is one part or type of dialectic.29 As we shall shortly see, there is no inconsistency. As Aristotle understands these initial requirements for peirastic any premises which satisfy them are guaranteed to be endoxa and, thus, premises for dialectical arguments, albeit of a special type. To better understand what peirastic arguments involve, it is worth looking closely now at the passages in Sophistical Refutations 1 1 where Aristotle introduces the important claim that the dialectician who is doing peirastic does his testing on the basis of a certain kind of knowledge. He describes this knowledge there as knowledge of what is "common" (koina). The dialectician who does peirastic is, in fact, identified at the beginning of that chapter as the one "who takes into account the common things as they relate to the matter at hand." (17lb6) These common things are described in the chapter as things such that "the one who does not know them necessarily does not possess [scientific] knowledge." (I 72a26-27) This description clearly reintrodu­ ces a feature of one of the special requirements for peirastic dialectical premises introduced in Chapter 2 (that they are necessarily known by the professing scientist) and thus it makes it clear that these common things are the actual premises involved in doing peirastic. Aristotle

further describes the common things which peirastic dialectic uses in the following way: Peirastic is not the scient.ific knowledge (episteme) of any definite . . may deal with every subject. For each of the arts subject. Thus it employs some common th!ngs (koina) so that everyone., even the unlearned, makes use of dialectic as peirastic in some fashion. For everyo1_1e att�mpts to test those who profess knowledge, on a limited basis, and this basis is the common things; because people know these things themselves no l�ss than those �ho profess knowledge even if they . s�em to say qu1.te 1naccurat � things. So they all practice refutation since they . prac�ice a� unsk1l �ed form of the activity which dialectic . with skill, while the dialectician is the one who does peirastic by practices use of an art of reasoning. (I 72a27-36)

(29) W. D. Ross, Aristotle's Metaphysics, 1924, v. 1 p. 260 £., C. Kirwan, Arislolle's Metaphysics IV-VI , 1971, pp. 84-5, P. Moraux "La joute dialectique d'apres le huitieme livre des Topiques" in Aristotle on Dialeclic, G. E. L. Owen (ed.) 1968� pp. 288-9 n. 3, followed by Brunschwig (1984/5), p. 35. These writers also charge Aristotle w ith a further , inconsistency, that of describing peirastic sometimes not as a part of dialectic, or as distinct from dialectic but as' identical to dialectic. This charge is based in part on Aristotle's use in Topics V I I l.5 of the term peira, trial or testing, in describing dialectic quite generally. But the term peira, like the verb peiran, is an ordinary Greek word used in Plato, Aristotle, and elsewhere to denote any kind of trial or testing, including any type of dialectical examination. The term peiraslike, by contrast, is a phiiosopher' s term of art, not used even by Plato and introduced with a technical definition by Aristotle in SE 2. There is no reason to suppose that any example of dialectic which Aristotle would describe as peira must fit the technical requirements of peiraslikl. Secondly, these writers claim that at SE 1 1 172a21, Aristotle says that dialectic "is peirastike." But this just means, in context, that it is dialectic which is capable of peirastic testing not some special episllml. This does not mean that everything that dialectic does is peirastic, only that peirastic is a capacity of dialectic. Cf. 171b9 (with b4-5) and Melaph. 1004b22-26, where the same point is made.

What are these "common" things which everyone knows? One possibility, accepted by some interpreters, is that they are what Aristotle elsewhere calls the "common notions" or "common axioms" - the principles which hold true for all subject matters. (APo. 1 . 1 0 76bl I ff; Melaph. II.2 996b26 ff) I n fact, o n one manuscript tradition, the one preferred by recent editors, Aristotle goes on immediately to characterize these "common" things as "identical principles which hold true of everything." (1 72a36-7) This is a natural reading in view of the fact that Aristotle contrasts these things with others which are "unique" (idia) to a given subject matter and such as to determine "a distinct nature and kind" (phusin tina kai genos, 172a37-8). This is just the kind of language which Aristotle uses elsewhere to contrast the common axioms with the other first principles special to each science. (APo. I . 1 0 76a37 ff, 76bl 1 ff) But this cannot be what Aristotle means here. For one thing, Aristotle emphasizes the point that only "some common things" are employed by each art. (1 72a29-30) So he does not mean that they all hold true of everything. In addition, if he did mean this, we would be faced with the bizarre consequence that the peirastic dialectician only examines or refutes claims by testing whether they fit with the common axioms - that is, principles such as the law of non­ contradiction, the axiom of equals and the like. But Aristotle permits peirastic dialectical argument to proceed from any relevant endoxa not just from common axioms. He gives various examples in the Sophisti­ cal Refutations of such arguments. Earlier in Chapter 1 1 , for example, he offers the following example of an argument which appeals to something "common." Take the case where someone denies that it is better to take a walk after dinner [rather than before dinner] and gives as his reason Zeno's argument [that motion is impossible]. This is not a medical argument _ since the argument employs what is common. (l 72a8-9) The argument mentioned here is in Aristotle's view an eristical or sophistical argument, not a dialectical argument. (See 172a2 ff) But it is like peirastic ? ialectical arguments, he says, in appealing to what is

216

I

t:I

I

. ,

R. BOLTON

common. And in this case the common things are clearly not true or applicable in all subjects, but rather include, in the case of Zeno 's _ argument, such common and well known information as: To go a certam distance you must first go half-way. One might suppose that such information must at least be applicable in a number of areas, but this would, again, exclude from the dialectician's premises n � n: erous relevant endoxa. Aristotle himself calls it common because it 1s not "based on first principles which are unique" to the specific subject at hand, i.e., medicine, but rather on matters which can be adapted for use "against people in general who do not know what is possible and impossible in the subject at hand; since it will apply with such people." (1 72a4-7) He later explains why the common things can be so used with people in general: " because people know these things themselves n o _ less than the scientist." (1 72a33-4) Here Aristotle opposes what IS common to what is used in argument from proper first principles (i.e., in demonstrations) with those who have special scientific knowledge of those principles and describes it as what can be used specifically by and with all people who may lack such special knowledge, but nevertheless know these things. A similar contrast to this one 1s drawn earher m Sophistical Refulations 9 . There Aristotle distinguishe� scientists who _ can produce or evaluate refutations based on the first prmc1ples (archai) of a given science i.e., refutations by demonstration (1 70a20-34; 36-38); from dialecticians, who produce refutations based on "the common sources of argument for each art and discipline" (170a34-36) and examine refutations "which do not !all under any particular art" (i.e. are not demonstrative in character) but are "based on the common things," i.e., on the premises for proofs from endoxa (endoxoi syllogismoi) on any given subject (1 70a38-bl). These common premises which determine the dialectician's province are clearly not restricted to propositions which hold good of all, or many, subject matters but include any appropriate and relevant endoxa. This use of the term 'common' is found in the opening lines of the Rhetoric where it is specifically used to describe dialectic in both its questioning and answering modes. The passage is worth repeating. Rhetoric is a partner of dialectic. For both of them are concerned with the sorts of things which it is in. a certain way common to everyon.e to know and which require no specialized knowledge. Thus everyone in a certain way takes part in both dialectic and rhetoric. For everyone on a limited basis engages in examination and in submitting to argument; and also in defending himself and in accusing others. (1354al-6) Another passage, a few paragraphs later, develops this point: In dealing with certain people, even if we possessed the most accurate knowledge (episti!me), it would not be easy to persuade them by argument� base� o.n this �nowle?ge. �or argume.nt . a�cord1n� to .kno�­ ledge (eptslimi}is 1nstruct1on (dtdaskalia), and this is impossible in this

THE EPISTEMOLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

217

case. Rather it is ne �essary t ? con�truct our pr�of� and arguments by use of the common things (koina), JUSt as we said in the Topics about ordinary discussions (enteuxeis) with people in general. (1355a24-29) In the discussion of enieuxeis to which Aristotle refers, in Topics 1.2, he does not use the term koina but rather, as parallel to it, "the opinions of people in general." (tas ton poll6n doxas, 101a31) So it is certain common opinions, whether they hold true of one or many or all subjects which constitute the "common" things as they are understood here. The language of these passages in the Rhetoric so closely parallels that in the Sophistical Refutations that there seems little doubt that the thought of the two is essentially the same. So the term 'common' in these passages does not designate what is metaphysically common, i.e., true of or applicable to all or many subjects, but rather what is epistemically common, i.e., what is intelligible even to common people, those without specialized knowledge of principles. Given this, it seems clear that we should not follow recent editors in accepting the manuscript tradition on which Aristotle describes the common things which dialectic reasons from as "identical principles which are true of everything." Rather, we should prefer the alternate manuscript tradition on which Aristotle says simply that "there are many of these [common] things in each area . . . so that it is possible on the basis of these things to engage in peirastic testing on every subject." (1 72a36-9) On this reading, Aristotle makes it clear that peirastic will pay special attention to those commonly understood things that hold true in the particular area under investigation. When he goes on to contrast the whole class of common things with things which are "proper" (idia) to a given area (172a38) he does not mean that particular common things do not specially concern some particular area but only that collectively the common things are not connected with "a distinct nature and kind (genos) ." (a37-38) This is how he argues that peirastic (and eristic) argument cannot be demonstration, from "first principles" ( l 72bl-4), since this would require that the common things do collectively determine a distinct kind.30

(30) The popularity of the alternate manuscript reading is no doubt influenced by the fact that in certain passages, such as Rhetoric 1.2 1358al ff, Aristotle draws a contrast between those dialectical or rhetorical arguments which draw on material which may be applied in common (koinon, 1358a l2) to all or many subjects and those which draw on matters which are unique (idia, a 1 7) to a particular subject. But this is not the same contrast as the one which Aristotle draws in the above passages. I n the above passages Aristotle uses the term koina ("the common things") to describe and delimit the materials for dialectical, or peirastic, reasoning quite generally. In Rhetoric 1.2 1358al ff, by contrast, the term is used to refer only to one type of basis for a rhetorical or dialectical syllogism by contrast with another (al 0-21, cf. 1.3 1359a16-26). In fact, Aristotle claims there that most rhet?rical syllogisms are based not on the koina, as he uses that term there,

'

I:

I I

i . I

THE EPISTEMOLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

It is easy enough to see now why it is that peirastic argumecyts always count as dialectical arguments, as those two types of arguments are defined in Sophistical Refutations 2. The premises of peirasbc arguments are the "common" things, things which are generallT accepted and, indeed, things which everyone knows, both the learned _ and the unlearned (I 72a27-36). Thus, the premises of peirasbc _ arguments are endoxa, by the definition of that terr;: m Topics I . �; More than that, since everyone accepts them, they are most endoxon or "as endoxon as possible" in the absolute sense. This explains wh� Aristotle, in Sophistical Refutations 2, can say that the quest10ners premises in peirastic arguments must be "believed by the answerer," when this is not all a requirement of dialectic m general. . Everyone accepts the premises of peirastic arguments. It al �o explams why, m Chapter 34, he can show a special interest m peiraslic as a proced �re m , which both questioner and answerer proceed from what is most endoxon" or "as endoxon as possible." Peirastic is, without quahfica-

as geometry but is rather an art one may possess even without having specific (scientific) knowledge" (1 72a22-3), Aristotle offers the following argument.

I

tion, a procedure of that sort. . . As we noted earlier, Aristotle takes it that one can legitimately criticize a dialectical argument not only where the premises are not endoxa, or where they are Jess endoxon than the conclusion, but also where the premises are false. (Topics V I I I . I I 1 6 1 b l 9 ff at t.21 ; 12 162bl4. The criticism can be overridden where the argument is used as a reduclio to establish the falsity of a premise 162bl6-22) It is clear that Aristotle understands all of these modes of criticism as evaluations of an -argument as a piece of dialectic, not from some �xter�al non-dialectical perspective. But it is a general puzzle how dialectic, which Aristotle always characterizes as proceeding from the point of view of standing opinion (kala doxan) rather than from the pomt of view of truth (kal' aleiheian), can be in a position to evaluate premises as true or false. (Top. I . 1 4 105b30-31 , APo. l . 1 9 S l b l S-23) Since Aristotle describes peiraslic dialectic as proceeding from what everyone knows and what even professing scientists must know, it is clear that for peirastic at least there must be some way of guaranteei'!g that its _ premises are true. If we can understand how this is accomplished this may contribute to the resolution of the general puzzle. . . There is one further important descnptton m Sophislzcal Refuta­ tions 1 1 of the common things which are known and understood by all and so serve to make genuine peirastic dialectical examinat�on possible. To show why peirastic dialectic "is not the same sort of thmg

. ,

•

219

R. BOLTON

218

but on things which are special to the subject at hand. {1358a26-28; cf. Ill.22 1395b3096b18} In SE 9 and 1 1 , however, all peirastic examination is based on koina, as the term is used there. (170a34-bl ; 171 b4-7, I72a27-bl ) So the two uses of the term are quite different.

It is possible even for someone who does not know the subject [scientifically] to test another who does not know the subject, providing that the latter grants things not based on what he knows [scientifically] or, in particular, on the special principles of the subject, but on things which are consequent on [the special principles] which are such that, though knowing them does not prevent one from not knowing the discipline, still one who does not know them necessarily does not know it. (l 72a23-27) What are these things different from but "consequent on" (la hepomena) the special principles of a science or other discipline such that in order to know the science or discipline you must know them as well as

the special principles? In the case of a science, the answer is clear; they must be the things explained by the special principles, the objects of the " knowledge that" which precedes the " knowledge why" provided by the principles. (APo. I I . 1 -2) This is all that the scientist needs to know in addition to the principles to have full knowledge of the science and these things are, of course, in a very straightforward sense "consequent on" the special principles. If this is correct then it is quite easy to see how peiraslic dialectic can play a crucial role in the process of discovery and j ustification i n science. If the "common" things, the special set of generally known endoxa, which peirastic dialectic argues from do make up at least some of the things which must be explained by the principles of a science then one can easily see how they can be used to rule out any proposed theories which are incompatible with them and how dialectical inquiry in a given area could serve to collect them together as a necessary prelude to the search for correct explanatory theories. But, as we have seen, these common things, the things which peirastic dialectic particularly argues from, are described not only as things consequent on the special principles of a science but also, later in Sophistical Refutations 34, as things which are "most endoxon." How can it be that there is this correspondence among these classes? Why should what is most endoxon turn out to constitute at least a part of what we must be able to explain by appropriate scientific principles? And why should either of these things be what is known by any scientist or commonly known (the koina) on the topic in question and thus true?

1 1 . DIALECTIC IN SCIENCE: A STANDARD VIEW At this point, it will be useful to consider the few attempts which have been made in the literature to say what it is that makes dialectic suitable as a procedure for j ustification in science or elsewhere, to see

220

I.

I i

II · . II ,,

.1

R. BOLTON

THE EPISTEMOLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

what they may offer in answer to these questions. One approach is suggested in the influential work done by G. E. L. Owen on Aristotle's . method of dialectic. Owen did not attempt to work out m any general or detailed way what the epistemological value is of the appeal to · endoxa. But he did argue that an appeal to endoxa or legomena (the things which are said) "may be an appeal either to common belief about matters of fact (e.g. EN I . l l l l0l a22-24) or to established forms of language (e.g. ibid. V I I . I . l 145bl9-20; 2 ! 146b4-5) or to a philosophical thesis claiming the factual virtues of the first and the analytic certainty of the second (e.g. ibid. 1.8 1098b l2-18)."31 Owen did not elaborate on ,, what the "factual virtues" of "common belief about matters of fact might b e for Aristotle. But he suggests here that the epistemological value of the appeal to "established forms of language" derives from the fact that they have "analytic certainty." In the same vein, Owen argues that when, for example, Aristotle a.rgues dialectically in the Ethics against the suggestion that practical wisdom can be coupled with incontinence on the ground that 'no one would say this' (l 146a6), what he is asserting is "not that it happens to be false but that given the established use of the words it is absurd." He claims further that in his discussion of incontinence, Aristotle appeals to certain endoxa in order "to show a priori that there is no use for the expression 'doing what is wrong in the full knowledge of what is right in the given circumstances' ."32 Owen does not elaborate on these claims at any length; and it would be easy, in the light of current doubts about the coherence of the analytic-synthetic distinction and its value for epistemology, to ignore his proposals. After all, there is no direct evidence that Aristotle recognized a class of analytic statements. He has no term for our 'analytic,' and he never explicitly attempts to justify belief in any class of statements on the ground that they hold simply in virtue of linguistic conventions which fix the meanings of terms and . so can be known a priori. But there are various reasons why the suggestion deserves consideration that appeal to endoxa is epistemically . significant, in some cases at least, because the endoxa reflect established forms of language and are thereby analytic and known a priori with certainty.33 If it were the case that peirastic dialect in particular did argue from analytic premises which reflect a priori conventions of

language that would in fact explain certain things that Aristotle says about peirastic. It would explain why the premises of peirastic argument are commonly known and accepted by everyone, and also why they are true and thus capable of being used to decisively refute proposed scientific principles which are incompatible with them. However, it would not explain one other crucial claim that Aristotle makes about peirastic premises, namely that they are things which are "consequent on" and explained by the relevant scientific principles. Why should the analytic truths have this status? It might be suggested in defense of Owen on this point that Aristotle wants to argue that the body of common or accredited opinion on a subject fixes the reference of the name of the subject in such a way that it is a logical presupposition of successful reference to that subject that most, and the most intelligible parts to us, of that body of opinion are not false of that subject or, more strongly, are explained by the basic principles of that subject.34 In his own discussions of reference Aristotle does suppose that there is some most intelligible opinion which at least sometimes has this kind of reference fixing status, namely the opinion expressed in the "account of what the name signifies."" Interestingly enough, this type of opinion can play for Aristotle the role of being one which appropriate theoretical principles must explain on pain of failure of reference and hence failure of truth for those principles. To take a familiar example from Posterior Analytics I I .8-10, Aristotle supposes that our belief that 'thunder is a certain noise in the clouds' serves as the belief by grasp of which our thought is fixed on to the real thing thunder in such a way that that fact must be true, and explained by the fundamental principles of meteorology. (93a21-9 with 94a7-9) A similar example from Paris of Animals I I .3 explicitly employs semantic terminology. It is clear that blood is hot in the following way, namely as something in the being of blood. Blood is said to be hot in the same way as something would be if we were to signify by its name that it is seething hot water. But blood is not hot in respect of its permanent substance (hupokeimenon). So in one way it is hot essentially (kath haulo) in another way it is not. For heat belongs in our account of it in the way in

(31) G. E. L. Owen, "Tithenai la Phainomena," p. 1 17. (32) P. 1 15. (33) Owen also makes the claim that "endoxa rest on experience even if they misrepresent it. If they did not Aristotle could find no place for them in his epistemology." (p. 1 17) He does not explain, however, what it is for endoxa to "re�t" on experience. If this is to be compatible with his claim that endoxa are often analytic and known a priol"i, it presumably cannot mean that all endoxa depend evidentially on or are known by experience.

221

(34) M. Nussbaum apparently has in mind a version of this line in saying that the ultimate justification for the appeal to endoxa or appearances is that only within the circle of appearances can we "refer" at all. (The Fragility of Goodness, 1986, p. 257.) See below for further discussion of her views . (35) APo. 11.10 93b30. Various different accounts have been offered in the literature of what sort of form and content this crucial reference fixing opinion has in Aristotle's theory. For discussion see R. Bolton, " Essentialism and Semantic Theory in Aristotle," Phil. Rev. 1976. The differences among these accounts do not affect the point in question here.

222

[!

' ' ' '

• !

.1 1'

,

,

:1

R. BOLTON

which white belongs in our account of the white man; but insofar as blood (pathos),36 it is not essentially hot. takes on heat as an accidental affection ,. 0) 649a13-2 cf. (649b21-27, Here Aristotle makes it clear again that in the account of what we · signify by a name a certain group of the features which we take to belong to what the name denotes will be included. He also makes it clear that the features which are included in the account of what we signify by a name may be features naturally present and whose presence is explicable. For instance, the mention of heat is included in the account of what we signify by the name 'blood,' even though heat is not a part of the fundamental essence of blood. Nevertheless blood is naturally heated up, particularly by the heart, in the course of its production for the sake of nourishment and growth, and it is in looking for the explanation of this that we find the real nature of blood. (650a2 ff, Juv. et Sen. 469b l2 ff) This provides a good illustration of how Aristotle's scientific inquiries standardly focus on looking for explanations for opinions with this kind of reference fixing status. But it offers only limited support for Owen's proposal at best, even putting aside the question whether these opinions count for Aristotle as items of a priori knowledge. Since Aristotle only gives such reference fixing status to single select opinions on a given subject, and never to the body of accredited opinion, or of most accredited opinion, as a whole, the role in justification which he accords to peirastic premises in general cannot be explained along these lines. Related problems arise from the fact that the appeals to "established forms of language" which Owen himself cites from the Ethics, do not seem to introduce analytic theses. The first of these is: "People are incontinent in respect of pride and horror and gain." The second is: "Some people are incontinent without qualification." In neither case does Aristotle regard the fact that "we say" these things as guaranteeing without further argument that they are so. He takes it as necessary to justify this pair of related theses (in Nfromachean Ethics V I I .4) and his justification for them does not draw solely on premises which are based on established forms of language. The justification depends crucially on the claim that some things that are a source of pleasure are necessary for life - such as food and drink; some are desirable for their own sake but not necessities - such as horror. (l 146b23 ff) On the basis of this he justifies the view that incontinence in respect of horror occurs and also the view that incontinence without any added qualification occurs (when a desire for some necessity overmasters a judgment of what is best). But this basis for Aristotle's (36) See GC 1.4 3 l 9b8 ff on the relation between hupokeimenon and pathos.

'

I,

THE EPISTEMOLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

223

argument is hardly analytic. Nor does Aristotle try "to show a priori," as Owen claims, that no one can do what is wrong "in the full knowledge of what is right in the given circumslahces. " In the familiar passage to which Owen refers Aristotle does claim that for the incontinent person to know in this way "would be astonishing" (lhaumaslon, l 147a9) or "would be taken to be strange" (deinon l 1 46b35). But this does not establish that Aristotle takes it to be evident a priori that the incontinent person does not know that he is doing what is wrong in this way. For one thing, Aristotle knows that strange and astonishing things do turn out to be true (e.g., that the diagonal of a square is incommensurable with the side, Metaphysics I.2 983a 15-1 7).37 More importantly, he takes it as necessary to argue for another account of the knowledge which the incontinent person has as an alternative to this one and, more significantly, to argue against attempts to defend the astonishing result. In both cases the argument depends clearly on a posteriori matters of fact. In the former case ( 1 1 47a l 1-18) Aristotle appeals to what "we see" in the case of people who are asleep, mad or drunk. In the latter case his argument continues as follows: The fact that someone [who is incontinent] makes claims from his own knowledge is no indication [that he is genuinely activating this knowledge]; for th?se who are in such conditions [of being asleep, mad, or drunk - which are like incontinence] even recite scientific demonstrations and verses of Empedocles. (1 147a18-20) Here Aristotle evidently uses a claim which he has just established - namely that the effect of strong appetites on knowledge in the incontinent person is on a part with the effect of strong drink on knowledge in those who are drunk - to undermine a powerful argument designed to show that incontinent people do sometimes fully well know that what they are doing is wrong. This argument is that they sometimes openly and correctly say such things as: " I know what I'm doing is wrong. "38 Aristotle's reply is in effect this: "Saying correctly that one knows something, even with apparent conviction, is no sure indication of genuine activation of one's knowledge of that fact in case one is asleep, mad, or drunk. In such a state saying something

(37) See the passages in Bonitz, Index, under thaumadzein. (38) Aristotle's phraseology here is unusual - legein tous logous lous apo tes epis­ temes. (1147al8-19) This means at least that the claims about good and bad which incontinent people make may express knowledge which they actually have, as in the use of apo epistemes at Metaphysics I.4 985al 6 . But the emphasis in the added use of the article ifs - "from his knowledge" - suggests that the understood verb with this phrase is not simply gignesthai (come) but rather a verb such as lambaneslhai (are taken) which would indicate that Aristotle has in mind the affirmation of certain things selected by the incontinent person as things which are known. In this event Aristotle is addressing himself to a tougher objection to his own view than is usually supposed. ,

I 224

225

R. BOLTON

THE EPISTEMOLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

coherent with conviction is not even a sign of genuinely understanding what one is saying. And incontinence is like these states." This argument is clearly not a priori. It is based on, among other things, an appeal to our common experiences when asleep or drunk. But the argument is a crucial part of Aristotle's defense of the claim that the incontinent person does not operate "in the full knowledge of what is right in the given circumstances." This makes it quite unlikely that Aristotle intended to defend this claim on _a priori grounds. It is however true, as Owen claims, that Aristotle does dismiss the suggestion that practical wisdom can be coupled with incontinence with the retort: "That is absurd . . . not a single person would venture to say [that]." ( 1 146a5-7) But does Aristotle think it absurd simply in view of linguistic conventions or "the established use of the words," as Owen says? Aristotle's own appeal here is in effect to the fact that absolutely everyone would reject the claim with the strongest conviction. He introduces no further qualifications. This would, however, be an adequate basis for rejecting the claim out of hand, as Aristotle does, if on the rules of the method which he is using here the strongest basis for rejecting (or accepting) some thesis is simply and without further qualification that it would be unequivocally rejected (or accepted) by everyone. This, as we have seen, is just what the rules of dialectic and peirastic require. Given this, it seems that Aristotle is here simply making a standard dialectical move without illuminating what it is that provides the ultimate epistemic backing for such a move.

opinion on which this [position] presumes." The endoxa are, rather, a set of convictions which, at least when properly refined, "hang together to constitute a world" in such a way that, ultimately, the accord of any beliefs with these convictions is "criteria! of their truth."" A typical development of this approach begins with the suggestion that, for Aristotle, the endoxa are simply one sub-class of the phainomena (the appearences) where the phainomena are to be understood as constituting the world as it exists for us. The phainomena, so understood, may of course conflict but in resolving any conflicts, on this view, we must square our results with the phainomena and show that these results preserve the phainomena as true, or at least that the greatest number and the most basic are preserved. In deciding which are most basic or authoritative, on this proposal, Aristotle considers our actual practices in justification to see, in different areas, what judges we in fact rely on. This reliance, however, does not itself require further justification for him. The reason for this is that "appearances and truth are not opposed, as Plato believed they were. We can have truth only inside the circle of appearances, because only there can we communicate, even refer, at all." This does not absolutely prohibit the introduction of new views but it does require that any new view draw on and preserve beliefs which are "deeper" than the ones which the new view forces us to give up - depper in that "the cost of giving them up would be greater, or one we are less inclined to pay." Given this, we can see how certain endoxa, those which are deepest or most endoxon, can be in a way a priori, by being "unrevisable, relative to a certain body of knowledge" or even so basic that they "cannot significantly be questioned at all from within the appearances, that is

12. REVISION OF THE STANDARD VIEW As noted above, the current philosophical climate is less hospitable now than it was when Owen wrote to appeals to what is analytic and know a priori on that ground; and this has led to greater caution in attributing appeals to the analytic to historical figures. So il'is perhaps unsurprising that more recent writers working within Owen's framework have shifted ground somewhat and argued that while the force of dialectical argument does not depend ultimately on the appeal to what is analytic and in that way absolutely a priori, it may involve an appeal to what is a priori in a different way. The suggestion has recently been made in various quarters that the value of the appeal to endoxa derives from the fact that the endoxa for Aristotle are not to be understood as a set of convictions which may or may not correspond to some belief-independent objective reality where independent evidence may be sought for their truth or falsity. "Aristotle does not make the clean separation between evidence and

(39) The quoted phrases are from M. Burnyeat, "Good Repute," J. McDowell, "The 'Role of Eudaimonia in Aristotle's Ethics" (in Essays on A ristotle's Ethics, ed. A. Rorty, p. 372) and M . Nussbaum, The Fragility of Goodness, p. 242. The description of the development of this position is mainly drawn from Nussbaum (pp. 240-258). A similar, t� ough not obviously identical, development is suggested by Burnyeat and by McDowell's . d1scus�1on (pp. 369-373). McDowell's remarks are, however, made in the course of reflections on Aristotle's method in ethics, and it is not clear that he would take them to appl?' in other �reas. Burnyeat and Nussbaum take the method to be universally apphca ?Ie �or Aristot�e. In T. Irwin, Arislolle's First Principles, it is argued that in his later scientific works (including the Metaphysics and De Anima) Aristotle introduces a new form of "strong dialectic" which draws on endoxa which are known a priori in a Kantian sense. They are �ot constituitive or reality but necessary preconditions of the possibility . of kno"".' ledge (or s1gn1ficant thought) abo'it reality. Since Irwin, unlike other writers, is not trying to account for the role of dialectic in science, as dialectic is understood and desc�ibed in the Organon, but only as dialectic is later used (but never described), his views require separate treatment. Reasons Cor doubting that Aristotle introduces in the Melaphysics a new form of dialectic, or uses a form which appeals to this sort of a priori knowledge, are given in "Aristotle's Conception of Metaphysics as a Science" (forthco­ ming).

227

R. BOLTON

THE EPISTEMOLOGICAL BASICS OF ARISTOTELIAN DIALECTIC

to say the lives and practices of human beings, as long as human beings are anything like us."40 In considering the merits of this type of view it is important to keep in mind that, as developed in the literature, the view is not always explicitly offered as an account of why it is that the method of dialectic, or its peiraslic part, is a useful and necessary tool to use in science or elsewhere but as an account of Aristotle's views on the proper method for the justification of belief quite generally. If the proper method is taken to be dialectic then these will, of course, come to the same thing. But even if it is not, in crucial respects the rules of the method as described here coincide with the rules of the method of dialectic and peirastic as we have seen it laid out by Aristotle in the Topics and elsewhere. In particular, the rule that where the things which are apparent to us conflict we are to prefer the ones which reflect what we as a group are more inclined, or more strongly inclined, to accept is just the main rule of dialectical method as a tool for inquiry. Peirastic in particular argues exclusively from what is most endoxon in an absolute sense, i.e., what is most apparent to us. So if the justification offered for this procedure were indeed Aristotle's then that would enable us to see what the basis was for his confidence in dialectical method and in peirastic in particular. This view has many of the same advantages in accounting for what Aristotle says about dialectic and, in particular, peirastic as those listed earlier for Owen's view. If peirastic argues ultimately from premises which are constitutive of reality and impossible for us to intelligibly question, then that would explain why its premises are true and known and accepted (at least implicitly) by everyone. Moreover, these advantages are achieved on this view, without the difficulties which come with the introduction of the notion of analyticity in interpreting Aristotle. However, the view would still not account for how it is that the premises of peirastic in particular are consequent on, and explained by, the first principles of the various sciences. In addition to this, there are reasons for doubting whethe� this is an adequate account of Aristotle's views on method quite generally. To begin with, there are the two respects noted earlier in which Aristotle's views on method in science go counter to the main rule given above. Aristotle takes it that in science the data of perception always take precedence over endoxa without reference to how widely held or how deeply entrenched those endoxa are.41 Secondly, he holds that

the.re is a procedure for the justification of first principles in science which goes beyond the matter of their ability to preserve or their inferability from what is most endoxon. In general, his doctrine is that such principles must exhibit and conform to what is most intelligible s1mp.ly or by nature by contrast with what is most intelligible to us. This reqmres, as is reflected in his discussion of peirastic, that these principles must be capable of explaining the appropriate endoxa on some subject rather than simply preserving them or being coherent with or derivable from them. In addition, these principles must be capable of explaining the relevant perceptual data whether these count as endoxa or what is ' most endoxon or apparent to us, or not. It may be responded to this that the main rule as described above can accommodate these points.42 First, it can accommodate the idea that in science perceptual phainomena always take precedence over endoxa if one of the things that is most apparent and endoxon to us just is the rule that in science perceptual data always take precedence over endoxa. Against this suggestion, however, a number of things can be said. To begin with, when he says we must give the top preference to perceptual data in science, Aristotle does not say or suggest that the reason for this is that people in general adopt this rule. Moreover, Aristotle nowhere says or suggests that people in general in his day even held such a rule. This rule was almost certainly only held by a very small minority of philosophers and scientists at best 43 and it seems unlikely that people in general held any clear views on �cientific method in Aristotle's day. Science was after all a quite new affair and the vast maj ority of people were no doubt entirely without views on the matter of scientific method. Aristotle does claim, as we have seen, that people m general go with the experts where they have no contrary views of their own. But if most experts on scientific method rejected Aristotle's principle in his own day it follows that from the point of view of the judges we trust or what is acceptable "to us" Aristotle should have rejected it. Another, deeper, question is how such a rule could ever be one which it was rational to adopt given Aristotle's general view on the proposal in question here. On the main rule proposed, Aristotle's view 1s not that we should follow the rules of method which are apparent, or most apparent, to us, but that we should follow whatever is overall most apparent to us with no special standing given to rules of method as such. Given this general principle how could we, or Aristotle, ever adopt the specific rule that in determining what should be accepted in

(40) Nussbaum, pp. 257-8, 254. (41) Nussbaum (pp. 268, 274, n . _21) takes the perceptual phainomena to be one subset of the phainornena, understood in general as what we think or what we say. But there is no reason to suppose that whatever careful observers perceive, as Aristotle understands this in passages such as APr 1.30, is always something that we believe or say.

(42) Cf. Nussbaum pp. 248-50. {43) We do not know of any Pythagorean .or Eleatic or Platonist, or of any Atomist or other post-eleatic pluralist who held this view. Certain medical writers may have held a view like this hut they constituted a small minority among theorists on scientific method. ,

226

:· '

11

I I

I

228

'

II

: 1.

•

I

H:11 I I

i' I,

I!

R. BOLTON

THE EPISTEMOLOGICAL BASIS OF ARISTOTELIAN DIALECTIC

science, perceptual data always take precedence over other appearances? On the general rule, the alternate specific rule that we should accept what scientists say, including their perceptual reports, where and only where it is compelling in the light of the total scheme of · what is apparent to us overall, is clearly a preferable rule. But if so, then there is no way on this proposal, of justifying the specific rule which Aristotle adopts - always to prefer perceptual data in determining what counts in science no matter what trouble this creates for our system of beliefs overall. It is if anything more difficult on the proposed general rule to see how to accommodate the specific rule that when it comes to scientific principles what is more intelligible simply, that is what explains the appropriate perceptual data (whether these are endoxa or not), takes priority over what is more intelligible lo us. The general rule is in effect that what takes precedence is always what is more intelligible lo us. How could it be that on the basis of this we decide to adopt the rule that for scientific principles what is less intelligible to us but more intelligible simply is to be preferred, or the rule that for scientific principles whether they are endoxa or not is irrelevant? (Posterior Analytics 1.6 74b21-26) It might be replied that this is in fact the way we do find it more intelligible to view scientific principles even when these principles conflict with other things that are more intelligible to us. But this is not in fact Aristotle's own view. As we have seen, he supposes that people in general are prepared to find special scientific views acceptable only when they do not conflict with other generally held views of whatever sort. Given this there is no way for him, or for us as he describes us, to move from the principle that we should always prefer what is more intelligible to us to the rule that in our scientific principles we should always prefer what is less intelligible to us but more intelligible absolutely. Even if the rule to prefer in science what is more intelligible simply to what is more intelligible to us is one acceptable "to us" if some specific scientific principle turns out ot be less acceptable to us overall than something which conflicts with it that we accept, we could not accept that principle, on the general rule, even if we' accepted the suggested rule of scientific method. Given these difficulties it is apparent that we cannot use the proposal offered to try to show what the value of dialectic, and in particular of peirastic, is for Aristotle. On this proposal, Aristotle regards the appeal to what is most apparent or most endoxon as ultimately an appeal to what fixes the limits of what is intelligible for us and thus cannot intelligibly be questioned or challenged. Peirastic is based on .the appeal to what is most intelligible to us. But since we have seen that Aristotle does not in fact view even what is most intelligible to us as ever, for that reason, beyond question or challenge by

refer�nce to perceptual data or to what is less intelligible to us but more mtelhg1ble absolute ly, this will not help us to see what the epistemologi­ _ cal basis of d1alect1c, or its value for science, is.

229

13. DIALECTIC IN SCIENCE: THE TRADITIONAL VIEW

By contrast with these recent approaches, which see dialectic as draw1 ng on the a priori, the most frequent traditional point of departure _ _ on this topic has been the group of passages where Aristotle talks about the relation between endoxa, or common opinions in general, and truth. 1:he . capacity to grasp the truth is the same as the capacity to grasp �ha.t 1s .hke the truth. At the same time, people have an adequate natural 1nchnahon toward the truth and so they turn out to attain truth for the most part. .Thus, the one who can aim skillfully [in argument] in relation to the truth 1s equally able to aim skillfully [in argument] in relation to the endoxa. (Rhetoric I. I 1355a l4-18) . . T�e study ? f �he truth is difficult in one way but easy in another. An �nd1?at1on of this is the fac.t that �hile no individual can adequately attain it still we do not all collectively fail. Rather each person has something of value t� say about the world and, though individually he contributes little or. nothing to �he truth, out of the assembly of all something significant arises. T� us, . 1nso �ar as the truth seems like the proverbial door which no one can miss, in this way the study will be easy. That we can have some grasp of the whole and not be able to grasp some part shows the difficulty of it. (Metaphysics I I . I 993a30-b7) Some of these views [about happiness] are voiced by many and from . long ago, others by a few renowned men; and it is reasonable that neither gr<:up is entirely mistaken, but right on at least one point if not most. (N1comachean Ethics 1.8 1098b27-9) . In all t�ese ma.tters we should try to get conviction through argument using t�e things which are apparently the case (la phainomena) as evidence aJ? d as illustrations. .For it is best i � everyone is seen to be in agreement with what. we are �01ng to say, or 1f not that all should agree in some way. (This th�y wrll do when. their views are modified.) For everyone has someth.1ng of his own to contribute to the truth, and it is from this that we must give some proof of what we say. For from the things which are said correctly but not clearly what is clear will be reached by those who . at every stage they exchange things which are more intelligible advance.' if for the Jumble of customary assertions. (Eudemian Ethics I . I 1216b26-35) These passages have been the ones most often referred to in attempts to account for Aristotle's confidence in the method of dialectic." In the first of these passages Aristotle offers the rather optimistic view that people in general have a natural propensity toward

. (4�)

See, among others, G. Grote, Aristotle (2nd edn., 1880) p. 273; W. F. R. Hardie, . Artsfotle s Ethical '[heory (1968) p. 38, Barnes, "Methods of Ethics," pp. 506 ff.

: I

.!

231

R. BOLTON

THE EPISTEMOLOGICAL BASIS OF ARISTOTELIAN DIALECTIC

the truth which results for the most part in the attainment of truth. Variants on this theme are found in the subsequent passages. These passages do not, of course, show (against what was argued earlier) that Aristotle thinks that the method of dialectic or peirastic is on its own sufficient for reaching final results in science or elsewhere. In the first passage Aristotle claims only that endoxa are for the most part true, and he explicitly points out in the second and third passages that while in some general way or on some point people may have made substantial progress toward the truth, in some particular area or on various other points they may have got nowhere at all. (Cf. Meta­ physics V I Il.3 1029b3 ff, Politics I I I . 1 1 128lbl5-21 , 35-39) Nothing here conflicts with the point that new perceptual data which are not endoxa can have higher epistemic standing than endoxa. Still, if it is the case that where people do have views those views are likely to amount to something, that might seem to offer a basis for using dialectic and peirastic as a part at least of one's standard method of inquiry. But do these passages even provide an adequate justification for that? For one thing, since Aristotle allows that at some times and in some areas the endoxa will not be reflective of truth, it seems that he must accept the conclusion that, sometimes, dialectic will not be useful in some specific scientific or other inquiry. But, if not, then does he have a way of telling us when it is useful and when not? In addition, these passages indicate why Aristotle might have thought it valuable to consider received views on some topic but what guidance do they offer on how to proceed when the received views conflict, as they often do? In particular how can they explain why one might want to place more weight on views which are more endoxon and more familiar lo us? Since this is the way dialectic and, in particular, peirastic proceeds if these passages do nothing to explain this they. do not show why in science (or elsewhere) we should make use of dialectic or, more specifically, of peirastic. The standard discussions of these passages do not address these problems and questions. Recent writers, in particular, have tried to deal only with the more limited question of why it is, in Aristo't le's view, that people more often than not come to hold true beliefs. The answer which they have offered is that, for Aristotle, (I) human beings are by nature intelligent or inclined toward the truth and thus naturally capable of coming to true beliefs. In addition, for Aristotle (2) nature does nothing in vain, i.e., natural capacities have a purpose which is for the most part realized - in this case the acquisition of truth. So (3) people for the most part come to true beliefs." Whatever the merits of this argument, it does not tell us how to proceed when people's views

conflict, so it does not help to justify the dialectical, or peirastical, way of proceeding in such cases.46 But what can we learn from the argument itself? Those who have offered this argument on Aristotle's behalf have been careful to point out that Aristotle himself does not use this argument, and to claim only that he is committed to its premises. The _ then 1s whether Aristotle is actually committed to the first questwn premises? He commits himself to the first premise of the argument, as well as to the conclusion, in the passage from the Rhetoric quoted above. But what of the second premise? Does Aristotle believe that natural capacities or inclinations are for the most part realized in the manner required here? That is not clear. Stones have the natural capacity to fall toward the center of the earth. But does this mean that all or most stones at some time actually realize this capacity? No, since _ , certam cond1t10ns must be met for the capacity to be realized (e.g., a stone must be placed in a medium less dense than it to fall) and it is not clear that these conditions are at some time met, or that Aristotle thinks that they are, for all or most stones. (Physics I l .8 199b!8, 26; Meta­ physics IX.5 1 047b39 ff, 7, 1048b37 ff) This point can be urged even more strongly for the case of intelligence since intelligence is, for Aristotle, not merely a natural capacity but a "rational capacity," i.e., one which it is within the power of an agent to exercise or not depending on, among many other types of conditions, the agent's overall . mot1vat10nal state. (Metaphysics IX.5; cf. On Dreams 2 460b l l ff) So Aristotle cannot, and would not, accept the generalization, in premise 2 abov.e, that natural capacities, including intelligence, are for the m �st part realized, in the way required for reaching his concluswn. He would feel obliged to show that the appropriate co ? di�ions, motivational and otherwise, are met in a way adequate for bndgmg the gap between the first premise and the conclusion. If we can understand how he would do this then we can perhaps understand how he would deal with the questions and problems for his views and

230

(45) See Barnes, ibid.

·

(46) J Cooper (review of The Fragility of Goodness, Philosophical Review 1988, pp. 553-4) has argued that "it stands to reason" from Aristotle's perspective that those who s�em to be acute and who exercise their intelligence more, namely "the wise'', are more hkely to come to the truth than ordinary people. Thus for Aristotle their views are preferable in case of conflict. (Cf. p. 551) However, this is not Aristotle's own rule in p�irastic, or dialectic generally; if it were dialectic would lose its required effectiveness . with people 1n general. (Top. 1.10 I04a l l-12) Moreover, in fact, as Aristotle knows, the . special views of "the wise" tend to be massively conflicting. This means that the views of lh � wise cannot in fact be for the most part true but rather are mainly false. Since Aristotle never says without qualification that they are mostly true but only says this abou t the vi�ws of peo �le in general, we lack the necessary support for attributing to him _ _ the (1mplaus1ble) doctnne that the views of the wise are more often, or more nearly, true than the common views of people in general.

THE EPISTE·M OLOGICAL BASIS OF ARISTOTELIAN DIALECTIC

how he would defend his claim that dialectic and in particular peirastic is necessary in scientific inquiry.

simply. Therefore, it is necessary to proceed in this manner: from what is less clear by nature, though more clear to us, to what is clearer and more intelligible by nature. The things which are evident and clear to us, in the first instance, are things which are rather jumbled up. Later, from them, their elements become known and the principles serve to separate these [jumbled up] things. That is why it is necessary to proceed from the comprehensive to specifics; for the whole is more intelligible in relation to perception and the comprehensive is a sort of whole, since it embraces many things as parts. (Physics I.I 184al6-26) It is advantageous to progress in stages to what is more intelligible, since for everyone learning proceeds in this way - through things which are less intelligible by nature to things which are more intelligible. Just as in the area of action our task is to start from what is good to each individual and make what is good altogether good to each, so it is our task to start from what is more intelligible to oneself and make what is more intelligible by nature intelligible to oneself. Now what is intelligible and fundamental to individuals is often barely intelligible and encompasses little or nothing of the reality of the thing in question. But nevertheless, one must try to understand the things which are altogether intelligible by starting from things which are hardly intelligible but intelligible to oneself and progressing, as we have said, through these very things. (Metaphysics Vll .3 1029b3-12) In these passages Aristotle begins by asserting, as he does in the Eudemian Ethics passage, that we must proceed by starting from what is more intelligible to us and moving to what is more intelligible simply. The initial reason which he evidently offers for this, particular­ ly in the Metaphysics passage, is the rather obvious one that we cannot start anywhere else than with what we understand. This by itself does nothing to answer the question as to why we are required, as the Eudemian Ethics says, to give some kind of proof of our final theoretical results, which may not be intelligible "to us," by reference to what is most intelligible to us. This question is particularly pointed in view of the claim in the Metaphysics passage that what is intelligible to any given individual may have little or nothing to do with the reality of the things in question. Aristotle does, however, go on to indicate in the Physics why we need to provide such a proof. He supposes that the elements and first principles which are the things most intelligible by nature are the elements and principles of the wholes which are most intelligible to us. This is to say that the data which make up what is most intelligible to us must on the whole be explained by the things which are more intelligible simply. This is the view we found also in the passage in the Sophistical Refutations where Aristotle characterizes the things in some area of scientific inquiry which are "common," i.e. commonly known and accepted, and thus used by peirastic, as "the things which are consequent on" the special principles of the subject. ( I I . 1 72a29-34 with a23-27) If this is so then it is easy to see why, as the Eudemian Ethics says, there must be a kind of proof of the principles by reference to common opinions and why he characterizes these opinions,

14. DIALECTICAL JUSTIFICATION IN SCIENCE The beginnings of an approach to this matter are to b � found in Aristotle's remarks in the passage from the Eudemwn Ethics quoted above. (I.6 1216b26-35) In that passage, Aristotle introduces . an elaborate j ustification for a certain methodological . recomm�ndat10n. The recommendation is that it is desirable to get convict10n (p1sils) m a way that relies on appeal to the phainomena o ; what appears so to us. His reason for that is that if we proceed m this way then our results will be in agreement with what everyone holds, either simply or "whe_n their views are modified." The reason which he offers to explam why it is desirable to be in agreement with what everyone holds is that everyone has something to contribute to the �uth to such an extent th �t "we must give some proof of what we say by reference to people s _ opinions, at least when they are properly modified. Aristotle does not say how he supposes this "modification" should take pi.ace. But the _ rarely used word which he employs here (melab1badzern) 1s prommently used also in the Topics to describe the dialectical procedure of rev1smg views by reference to what is "more endoxon" or "as en�oxon as possible." (VI I I . 1 1 16l a33-37, with b34-38 and V I I I .5-6) Smee the procedure which Aristotle is describmg_ here_ also mvolves reasomng from what enjoys the maximum consensus, 1t 1s likely that this 1s what he has in mind. Here, then, Aristotle explicitly introduces the claim we have been trying to understand, the claim that it is not simply a desirable or useful point of method to appeal to the common opinions, or the most common opinions, on a topic but necessary. He goes on to _ g1ve the reason why this is so. His reason is that by proceedmg m this way we guarantee that we move from what is correct though somehow unclear and jumbled up to what is clear and more intelligible and, so he must be supposing, it is necessary that we proceed _ m this way.. . What exactly is this way of proceedmg? Why 1s 1t necessary? And how does proof by reference to what everyone holds play. a role m this procedure? The doctrine which Aristotle develops here 1s expres­ sed in quite similar language elsewhere, notably m the methodolog1cal passage with which he opens his Physics and in a related passage m Metaphysics V I I .3. The natural procedure is to go from things which�are m? re i�t� lligible and clearer to us to things which are clearer and more in.te!hg1b�e . by nature. For what is intelligible to us is not the same as what is intelhg1ble

�

'i :

:

i

!

:1· !. 11

' '

iI

·l!1,.·

233

R. BOLTON

232

234

'1

ii

ij·'II·: ,,

,,

R. BOLTON

THE EPISTEMOLOGICAL BASIS OF ARISTOTELIAN DIALECTIC

in the Eudemian Ethics, as correct. But why do the principles have to be principles of and hence to explain such data. Why cannot mosl of our common opinions on some subject, including most of the ones which are most intelligible to us (and thus most endoxon), simply be wrong and · thus not in need of explanation by any theoretical principles? The reason which Aristotle goes on to offer in the Physics passage quoted above in defense of his claim that what is most intelligible to us must be explained by appropriate principles is that these opinions accord with "what is more intelligible in relation to perception." This fits with a claim which Aristotle repeats elsewhere: At different times different things are more intelligible to the same people � in the beginning the objects of perception, but as their knowledge becomes more accurate, the reverse. (Topics VI.4 142a2 ff) I mean by what is prior and more intelligible to us what is closer to perception, by what is prior and more intelligible simply I mean what is further from perception. The things which are furtherest from perception are, most of all, universals; the things which are closest are particulars; and these are opposite to each other. (APo. 1.2 72a l ff)

Aristotle's remarks about the need for dialectic in science with what he says in these passages. It also shows why dialectic has the limits which we have seen it has in scientific inquiry. Peirastic draws on what is in conformity with the most empirically well-justified information that as a group we have up lo now. But further observation by the individual scientist or others may yield new empirical data which show that some theory which is most empirically well-confirmed for us now, because it offers us the best explanation for the empirical data which we have amassed up to now, is wrong. This may occur whether or not these new data are or come to he endoxa or generally known, i.e. a basis for further dialectical or peirastic argument. This explains why Aristotle can say that in science we must try to account for the endoxa but also suppose that perceptual observations take precedence over endoxa in validating theories. This account of the value of dialectic also explains why dialectic cannot by itself provide us with adequate theoretical principles. It serves to draw on and thus aid us in collecting the empirically most well­ confirmed data which theoretical principles must serve to explain. But by itself it cannot determine the principles which do the explaining. This is why, nevertheless, it plays a necessary role in finding those principles. Since it serves to collect the most empirically well-justified information that we as a group have now it plays an essential role in our search here and now for first principles. For the best we can do here and now in our search for first principles is to find those candidates for first principles which, among other things, do the best job of explaining the empirically most well-confirmed information that we have now.

What is most intelligible to us, Aristotle thinks, always coincides with what is not closely related to what we perceive . Elsewhere he is quite explicit that the fact that all or most people accept something "gives it credibility as something based on experience. "47 He specifical­ ly says in the Physics that this is why first principles must account for what is more intelligible to us. If this is so then we can explain what the value of peirastic dialectic is for science in the following way. We have seen that peirastic dialectic is a procedure for the testing of claims by reference to what is most endoxon and most intelligible to us. As such1 in Aristotle's view, it is a procedure for the testing of claims by reference to those beliefs ' which we have which are most closely connected with information whi ch we have acquired by perception and for the rejection of whatever claims may conflict with this information. So peiraslic dialectic, in effect, turns out to be a procedure which draws on the information which we as a group are now warranted in accepting on the basis of what is most obvious to us from perception. This explains how Aristotle can describe peirastic as drawing on what everyone knows and how dialectic, in certain forms, can draw on what may reasonably be claimed to be true. This also provides a clear role for dialectic in scientific inquiry understood in the way in which Aristotle standardly understands it in passages such as Prior Analylics 1 .30, namely as a procedure which starts from the data of experience and proceeds to find principles which conform to and explain such data. It enables us to coherently integrate (47) Diu. in Som. 462bl4-16.

235

15. T H E MERITS OF PEIRASTIC JUSTIFICATION

On the account which we have now reached of the epistemological basis of peirastic dialectic that method turns out not to embody any general appeal to what is a priori. Nor does it embody, despite certain appearances to the contrary, a strong coherence procedure for the justification of belief. A belief is not justified in peirastic dialectic simply when it coheres well with the maximal consistent body of existing belief or anything of the sort. Peirastic dialectical justifica­ tion, for a given belief does depend on whether that belief can be ultimately defended by reference to those beliefs which are "most endoxon" i.e., most widely and firmly accepted by us. But even here it is not because those beliefs are most widely and firmly accepted by us that the belief in question is justified by reference to them. Rather it is because those beliefs bear a special relation to the data of

236 .

1 ·.·

I

i:

l: I I

R . BOLTON

B10LOGIE, LOGIOIJE £t MttAl'ffYS!OlJE

experience. Peirastic dialectical j ustification, for Aristotle, like justifi­ cation in science simply, turns out to give special priority (though not precisely the same priority) to the evidence of experience. This does not mean, however, that perceptual judgments are, for Aristotle, incorrigible. Though this is not the place for a full treatment of his views on this subject, it is clear enough that he does not take ordinary perceptual judgments to be in principle or in general unrevisable. (De Anima J I I .3 428bl8-25) This is compatible, however, with his general assignment of final epistemic authority in science to perceptual over theoretical beliefs. (De Caelo JII.7 306a l3-7) In peirastic dialectic this result• also holds. It follows from the doctrine that revision must always be made by reference to what is most intelligible to us, together with the doctrine that what is most intelligible to us is always closest to perception. In this respect there is a kind of foundationalist empiricist element in Aristotle's theory of justification both in science and in peirastic dialectic generally." This result raises a number of interesting questions. Perhaps the most important are these: How would Aristotle try to show that the more widely and firmly accepted something is the more likely it is to directly embody information acquired by perception? And, how would Aristotle justify the reliance which he places in "the testimony of the senses?" To answer these questions, an.cl other related ones, would require us to go into Aristotle's rather complex views on information acquisition; and this is not the place for that. Here it must suffice for us to see that, recent claims to the contrary notwithstanding, the old cliche that Aristotle is fundamentally an empiricist is, after all, in certain respects basically right."

SCminaire CNRS-N.S.F., 1987

I

' (48) There is not space here to consider the role of non-peirastic forms of dialectic in science. Aristotle is clear that periastic is the form of dialectic which is especially relevant in scientific, or other philosophical, inquiry. (Melaph. IV.2 1004b25-26, quoted above at the beginning) The account developed here can be extended to cover the other forms. (49) This is an extensively revised version of the paper presented at the conference on the Ile d'Olilron. The excellent penetrating comments of Jacques Brunschwig and Daniel Devereux on the original version served as the main stimulus for the revisions. They have led me to minimize discussion of controversial points whose defense is inessential to my main proposals. Brunschwig also kindly gave me at the conference a copy of his valuable paper (see n. 5) which I had not seen before; so I have rectified my failure to consider it earlier.

chez Alusrom

tditions du CNRS, Paris, 1990.

REMARQUES SUR LA COMMUNICATION DE ROBERT BOLTON JACQUES BRUNSCHWIG

Robert Bolton avait presente, au Colloque d'Oleron, une premiere version de l'etude magistrale dont on vient de lire la version definitive. C'est naturellement sur la base de la premiere version que j 'avais redige Jes quelques commentaires qu'il m'avait ete demande de presenter moi­ meme au Colloque, pour servir d'introduction a la discussion generale. Ces commentaires comportaient, selon la Joi du genre, quelques critiques et objections, portant sur des passages precis du texte de Bolton. Daniel Devereux, de son cote, presentait ses propres observations sur le meme texte, observations dont la convergence avec les miennes etait remar­ quable, et parfois meme surprenante. La version definitive du texte de Bolton� ici imprimee, n'a pu m'i\tre communique.e qu'assez recemment. Elle est encore plus riche et plus developpee que la version initiale, et a beaucoup d' egards, elle en est I?rofondement differente. Je disposais d'un temps limite pour reviser man propre texte ; ii m'a fallu deja reserver une bonne partie de ce temps pour reperer exactement les transformations de toutes sortes que !'auteur avait effectuees sur son texte initial. Ces transformations sont, d 'ailleurs, extraordinairement minutieuses et meditees ; et ce fut un privilege de pouvoir les observer de pres. Mais ii m'aurait fallu plus de temps encore pour adapter mes commentaires initiaux, de fagon tout a fait appropriee, a un texte si soigneusement retravaille. En etudiant avec tout le soin possible ce fruit des amours de Macintosh et de Penelope, j 'ai pu faire, pour ce qui me concernait personnellement, deux constatations principales. D'une part, la quasi­ totalite des passages (qu'ils fussent longs ou brefs) de la premiere version qui avaient fait l'objet de diverses objections de ma part ant purement et simplement disparu de la version finale, et ils ant ete remplaces par des developpements entierement nouveaux. D'autre part, une bonne

238

i

J. BRUNSCHWIG

REMARQUES SUR LA COMMUNICATION DE ROBERT BOLTON

part des adjonctions que le texte a regues contiennent diverses critiques qUi me sont nommement adressees, a propos d'un petit article que j 'ai consacre, il y a quelque temps, aux Topiques d' Aristote (« Aristotle on Arguments without Winners or Losers », in Wissenschaftskolleg zu Berlin­ - Jahrbuch 1984/1985), article dont il n'avait pas ete question, sinon tres allusivement, dans les debats d'Oleron. Pour desencombrer d' emblee la discussion de toute dimension agonistique, je voudrais d'abord dire quelques mots au sujet de cet article. Je n'ai nulle envie de me laisser entrainer a defendre ici Un petit texte dans lequel j e n'ai pas mis de grandes pretentions, et qui merite sans nul doute quelques-unes des critiques precises que m'adresse Robert Bolton a son sujet (je veux parler de celles contenues dans sa note 6, ou se trouve resolu tres habilement, du meme coup, un probleme difficile). Je suis mains d'accord, bien sfJ.r, avec la presentation generate qu'il donne de cet article ; elle fail silence sur les precautions que j ' avais prises pour y delimiter expressement mon sujet. Mais ce n'est pas ici le lieu d'en debattre. 11 n'y a pas de raison, me semble-t-il, pour que la discussion se deplace sur un texte dont le lecteur ne dispose pas, et dont le propos ne recoupe que marginalement le probleme qui fail l'objet essentiel de l'etude de Robert Bolton. En revanche, ii peut valoir la peine de suivre, au mains sur quelques points, les effets de nos discussions d'Oleron, et c'est ce que je vais essayer de faire ici. Observons d'abord, pour fixer le cadre de la discussion, que l'on peut distinguer, dans l'etude de · Bolton, et d'apres ses propres indications, une question centrale et une question prealable. La question centrale est celle-ci : quel est le role joue par la methode dialectique, chez Aristote, dans !'acquisition et dans la j ustification des resultats qu'il presente dans ses ouvrages scientifiques et philosophiques, , et surtout, quelle est pour lui la j ustification epistemologique de ce rOle ? La question prealable est celle-ci : qu'est-ce que c'est au juste que la methode dialectique en question ? Non sans raison, Bolton soutient que la diversite considerable des reponses apportees jusqu'a present a la premiere question, la question centrale, rend eminemment sOuhaitable une reprise attentive de la seconde question, la question prealable. Notons que le corpus pertinent n'est pas le meme pour les deux questions : la question centrale implique une enquete a travers les indications methodologiques contenues dans tous les traites. aristoteli­ ciens, alors que la question prealable porte essentiellement sur !'interpre­ tation de ce que Bolton appelle le « traite officiel d'Aristote sur la dialectique», a savoir les Topiques, au sens etroit du terme (sans les Refutations sophistiques) ou au sens large (avec les Refutations). Notons aussi que l'articulation des deux questions peut soulever d'assez nombreux problemes : ii n'est pas evidenL, ni hors de Loute contestation,

que les procedures qui, dans les traites scientifiques et philosophiques d' Aristote, peuvent etre qualifiees de « dialectiques » relevent toutes d'une seule et meme « methode dialectique » ; il n'est pas evident non plus que cette <• methode dialectique», au cas oil l'on en admettrait l'unicite, soit identique a celle qui est presentee theoriquement, et enseignee pratiquement, dans les Topiques. Mes commentaires initiaux, clans ce qu'ils avaient de critique, ne portaient pas, pour la plupart, sur les theses developpees par Bolton a propos de sa question centrale. Ils concernaient presque exclusivement la question prealable. Sous ce rapport, je l'ai dit, la nouvelle version de son texte me donne entiere satisfaction, et je ne puis que remercier Robert Bolton d'avoir bien voulu prendre si soigneusement mes remarques en consideration. Les sections 2, 4-6 et 9-10, qui constituent dans la nouvelle version l'essentiel du traitement de la question prealable, ont ete totalement ou largement reecrites, et la plupart de mes objections initiales he s'appliquent plus du tout a elles. On me pardonnera, je l'espere, d 1 avoir ete particulierement attentif a ces sections nouvelles du texte qui, au mains implicitement, repondent et reagissent a mes remarques. Quelques difficultes y subsistent, a mon sens, que l'on peut considerer comme des vestiges, plus ou mains reformes OU transformes, des difficultes qui m'etaient apparues initiale­ ment. Elles peuvent constituer la matiere d'une nouvelle discussion. J 'examinerai done successivement (et tres selectivement si l'on considere la multiplicite et l'ampleur des questions soulevees par !'elude de Robert Bolton) quelques themes qui, par eux-memes aussi bien que dans l'interpretation construite par mon interlocuteur, apparaissent comme etroitement lies.

239

J . Q UESTIONNEUR ET REPONDANT

Dans sa version initiale, Robert Bolton avail accorde fort peu d'importance au fait qu'Aristote, au mains a un niveau de Sa reflexion sur la methode dialectique et de sa description de cette methode, considere l'activite dialectique non seulement comme un dialogue, mais encore comme un dialogue regle OU meme reglemente, dans lequel Jes roles du questionneur et du repondant sont nettement differencies. 11 avait, par exemple, explique deux passages-clefs, oU Aristote annonce (Top. I I , 100a!8-21) et recapitule (Refut. Soph. 34, 183a37-b6) toute l'entreprise des Topiques, sans vouloir y reconnaltre la distribution des elements qui, de fagon pourtant hautement significative dans des passages de cette nature, se rapportent respectivement a la tache du questionneur et a celle du repondant. On ne peut plus reprocher a la nouvelle versio� de son etude d'avoir neglige cet aspect de la conception

240

J.

BRUNSCHWIG

aristotelicienne de la dialectique ; ce qu'il faut dire a ce sujet est dit avec clarte et exactitude dans des passages comme ceux que !'on trouve p. 188 (avec la note 6) et pp. 197-198. II est cependant visible que Robert Bolton tend ;\ minimiser la portee de cette concession, et ceci de diverses manieres au cours des diverses etapes que traverse la description, de plus en plus serree et approfondie, qu'il donne de la methode dialectique. Suivons la progression de cette description. Des la section 2, que Bolton presente lui-meme comme une « premiere approximation » (p. 187), « intentionnellement sommaire et a divers egards imprecise et incomplete » (p. 190), on voit apparaitre le concept de « justification dialectique», concept qui transcende la distinction du questionneur et du repondant, en ce sens que chacun des deux partenaires peut esperer conquerir pour sa these (traduisons ainsi. l'intraduisible « claim») une telle « justification dialectique ». Naturelle­ ment, Bolton n'ignore pas que cette etiquette unique doit recouvrir des marchandises differentes, selon que !'on considere la these du repondant (celle que celui-ci s'engage a soutenir et a defendre contre l'assaut dialectique de son adversaire) ou celle du questionneur (celle, contradic­ toire OU incompatible avec celle du repondant, que celui-ci entreprend d'etablir en obtenant ses premisses par Jes questions qu'il pose ;\ son partenaire). II doit done definir disjonctivement son concept de «justification dialectique», comme ii le fait p. 189 : « Une these peut etre dite ' dialectiquement justifiee ' seulement si, ou bien elle suit de fagon appropriee d'elements qui appartiennent a !'ensemble existant des croyances notoires ou accreditees, ou bien elle est compatible avec des elements de cet ensemble de croyances (en ce sens que sa contradictoire ne suit pas de fagon appropriee d'elements de cet ensemble) » (souligne par moi). Des expressions abregees de la meme disjonction se retrouvent, par exemple, p. 190 («[la dialectique] considere qu'une these est dialectiquement justifiee seulement dans le cas oil elle est, de fagon appropriee, compatible avec certaines croyances repandues, notoires ou accrectitees, ou impliquee par ces croyances ») ou p. 192 (« [les definitions dialectiquement adequates) peuvent etre inferees ;\ partir des endoxa OU d'un sous-ensemble approprie d 'endoxa, ou au mains sont compatibles avec eux )>) . Cette presentation des chases, sans etre inexacte, me semble avoir pour effet de dissimuler ou d'attenuer une difference cruciale entre les deux formes de la « justification dialectique ». La these du questionneur sera « j ustifiee dialectiquement » si et seulement si le questionneur parvient a en faire la conclusion d'un syllogisme dialectique, c'est-a-dire formellement valide et prenant pour premisses un sous-ensemble de I' ensemble des endoxa ; cette tache est en principe faisable dans le cadre d'un entretien dialectique unique et de duree limitee. Dans ce meme cadre, le repondant ne peut esperer rien de mieux que mettre en echec

REMARQUES' SUR LA COMMUNICATION DE ROBERT BOLTON

·

241

l'entreprise de son partenaire ; ii aura correctement defendu sa these, p, si et seulement si I' ensemble limite des premisses que le questionneur lui a proposees, et qu'il Jui a accordees, ne permet pas a celui-ci de parvenir validement ;\ la conclusion non-p. Dans de telles conditions, ii parait difficile de dire que p, la these du repondant, se trouve « justifiee dialectiquement » ; tout ce que l'on peut dire est qu'elle n'est pas, ou qu'elle n'a pas ete, dialectiquement disqualifiee ou in-justifiee. Pour passer de ce statut de non-injustification (ou, si l'on veut, de non­ infirmation) dialectique a un statut de justification ou de confirmation dialectique, analogue a celui que la these du questionneur est susceptible d'acquerir d'un seul coup, ii faudrait que la these du repondant soit soumise a une multiplicite d'epreuves dialectiques dont elle sortirait chaque fois indemne, chacune de ces epreuves se differenciant, non pas necessairement par la personnalite du questionneur, mais plutot par la selection des endoxa utilises dans la tentative de refutation de la these du repondant. Meme en supposant que le nombre des endoxa pertinents pour la discussion de cette these ne soit pas indefini, ii est clair qu'une entreprise de ce genre deborderait les limites de l'entretien dialectique, tel que Jes Topiques en definissent les regles du jeu. Une notion claire et univoque de «j ustification dialectique», notion vraisemblablement utile et meme necessaire pour l'etude de !'intervention de la methode dialectique dans les traites de science et de philosophie, ne semble done pas pouvoir etre aisement extraite de la description que donnent Jes Topiques de l'activite dialectique, en raison du rOle essentiel que joue dans cette description, 3. mon sens, l 'articulation des rOles de question­ neur et de repondant, et l'asymetrie des conditions faites aux theses respectives de l'un et de l 'autre. C'est pourquoi deux attitudes sont theoriquement possibles, me semble-t-il, par rapport aux Topiques : ou bien l'on considere que ce traite contient les elements pertinents qui permettent de repondre aux questions que pose l'emploi des procedures de justification dialectique dans les traites de science et de philosophie, et dans ce cas l'on est conduit a minimiser ou a neutraliser l'essentialite du rapport dialogique questionneur/repondant dans la methode dont traitent les Topiques ; c'est ce que fait dans ]'ensemble Robert Bolton, qui remarque avec bon sens qu'apres tout, lorsque Aristote procecte « dialectiquement» dans ses traiteS, ii fait lui-meme les demandes et les reponses (p. 1 99) ; OU bien !'on considere (comme j'ai plutot tendance a le faire) que Jes Topiques sont ecrits, pour l 'essentiel, dans la perspective d'un dialogue scolairement codifie entre questionneur et repondant, et !'on est alors conduit a penser qu'ils ne livrent pas immediatement (meme en tenant compte des quelques indications qu'ils contiennent sur les fonctions philosophiques de la methode) Jes elements necessaires a la comprehension du statut epistemologique de la methode dialectique dans les traites de science et

242

J.

BRUNSCHWIG

de philosophie. Il semble, en d'autres termes, qu'un hiatus subsiste entre la methode decrite dans les Topiques et la dialectique mise en reuvre dans les traites : des !ors, ou bien l'on tente de combler ce hiatus en considerant comme relativement accessoires les aspects strictement dialogiques de la methode des Topiques, ou bien on le laisse subsister en prenant ces aspects au serieux, quitte a admettre que la dialectique philosophique des traites ne se superpose pas exactement a celle des Topiques. Suivons maintenant le fil de cette question. Dans la section 4· de son etude, Bolton revient avec plus de detail, comme il l'avait annonce, sur la description intentionnellement sommaire qu'il avait d'abord donnee de la methode des Topiques (il s'agit maintenant, est-il
REMARQUES' SUR LA COMMUNICATION DE ROBERT BOLTON

·

243

On ne peut comprendre, selon Bolton, la valeur justificative du raisonnement dialectique qu'au prix de ce collapsus du couple question­ neur/repondant. En elfet, si l'on cherche it resoudre les cas (supposes) de conflit entre endoxa en permettant it chaque partenaire de s'appuyer sur tel sous-ensemble non-contradictoire d'endoxa qui convient 3 ses objectifs propres, le caractere arbitraire du choix de ce sous-ensemble reduira a fort peu de chose le pouvoir justificatif du raisonnement dialectique, par rapport aux conclusions qu'il est suppose etablir ou refuter. Meme si l'on imposait aux deux partenaires de s' entendre pour s'appuyer tous deux sur le sous-ensemble non-contradictoire maximal d'endoxa pertinents. la mise hors jeu des endoxa incompatibles avec ceux qui ont ete retenus diminuerait encore tres fortement la respectabilite epistemologique des resultats que l'on peut attendre de la method e ; dans cette hypothese, en ef!et, celle-ci ferait abstraction de certaines donnees pertinentes, pour la seule raison qu'elles ne s'accordent pas avec la majorite des autres. Ces difficultes relancent la reflexion de Bolton sur la «justification dialectique » dans une nouvelle direction. Dans un premier temps, on observe qu'il neutralise incidemment la distinction du questionneur et du repondant («ces questions . . . sont d'une importance toute speciale pour determiner quelle espece de justification epistemologique peut etre procuree it une these donnee (pour le quesiionneur, OU pour le repondanl, OU pour les deux) par un echantillon satisfaisant de raisonnement dialectique », p. 198 - souligne par moi) ; un peu plus loin, cette meme distinction se trouve mise entre parentheses, de fagon methodique et explicite. Non sans quelque paradoxe, apres avoir annonce qu'il examinerait de fagon detaillee « ce qu' Aristote lui-meme qui ne vise pas a l '« exactitude». Remarquons au passage qu'au prix d'une certaine incoherence, Bolton pense trouver dans les Topiques eux-memes les materiaux necessaires pour corriger leur manque d'<( exactitude». Ainsi, ils montre­ raient que la structure duelle de l'entretien dialogique n'est pas essentielle it la methode dialectique, puisque Aristote y recommande it l'apprenti dialecticien comme au dialecticien philosophe de s'y exercer tout seul (cf. les textes cites p. 1 98). Je ne crois pourtant pas que ces

I

r

'

I .

' ' '

245

J. BRUNSCHWIG

REMARQUES0 SUR LA COMMUNICATION DE ROBERT BOLTON

recommandations puissent servir d 'argument a la these de Bolton, car les exercices solitaires qu'Aristote y prescrit sont manifestement des exercices preparatoires, et rien dans le texte ne dit que celui qui s'y livre fasse deja de la dialectique ; on peut parfaitement penser qu'il ne fait · que se preparer a en faire, et qu'il n'en fera que lorsqu'il sera en situation dialogique. Quoi qu'il en soit, nous ne sommes pas au terme de l'elaboration progressive de la notion de «justification dialectique » que nous propos: Robert Bolton. Dans une derniere etape de son developpement, 11 soutient que l'on pourrait en atteindre la comprehension en considerant « la valeur epistemologique des resultats atteints par une voie qui se conformerait aux requisits conjoinls qui doivent etre respectes dans une discussion dialectique a deux personnes, a la fois par le queslionneur et par le repondanl, qu'il y ait ou non deux personnes reelles qui soient responsables de ces resultats » {p. 198 - souligne par moi). On remarquera qu'au terme de ce developpement, la notion de «justification dialectique » s'est sensiblement transformee : alors qu'elle etait, au depart de la reflexion de Bolton, definie disjonctivemenl (est dialectiquement justifiee une these qui satisfait soil aux regles imposees au questionneur, soil a celles imposees au repondant), elle est maintenant definie conjonctivemenl (est dialectiquement justifiee une these qui satisfait a la fois aux regles imposees a l'un et a l 'autre). Mais ii ne s'agit ici, j 'en ai bien peur, que d'un hommage rendu du bout des Ievres au couple dialectique du questionneur et du repondant : car cette definition conjonctive permet de considerer comme indifferente la question de savoir si les deux rOles s'incarnent ou non en deux interlocuteurs reels ; en outre, s'il etait vrai, comme Bolton a cherche a le montrer une page auparavant, que les taches du repondant et du questionneur (sous la condition, supposee remplie, de la possibilite d'un conflit entre endoxa) sont l'une impossible et l'autre triviale, on n'obtiendrait rien de plus substantiel en conjoignant les requisits qu'elles doivent remplir. N 'oublions pas, toutefois, que !'ensemble de ce developpement reste subordonne a l 'hypothese d'une possibilite de conflit entre endpxa. Sous ce rapport, on pourrait resumer comme suit le raisonnement de Robert Bolton : (a) les endoxa peuvent etre en conflit, et ils le sont habituelle­ ment sur les questions les plus difficiles et les plus importantes ; {b) quand ils le sont, le role du repondant devient difficile jusqu'a etre impossible, et celui du questionneur facile jusqu'a etre trivial ; (c) s1_ !'on veut comprendre que la methode dialectique decrite dans les Topiques puisse avoir une valeur justificative, a l'egard des enonces qui, dans les traites scientifiques, ·se presentent comme la solution des problemes sur Iesquels ii y a des conflits entre les endoxa, ii faut done debarrasser cette methode du vi\tement dialogique qui n'est que le vestige superficiel de son histoire anterieure.

II convient done d'examiner maintenant la premisse fondamentale de cette ligne d 'argumentation.

244

2. L A QUESTION DU CONFLIT ENTRE ENDOXA Qu'Aristote envisage la possibilite de conflits entre endoxa, c'est ce qui risque de sembler evident au lecteur d'ouvrages comme la Physique, la Melaphysique ou les Elhiques. Les recherches que contiennent ces traites sont regulierement introduites, on le sait, par des revues doxographiques oU se trouvent confrontees diverses opinions, opinions communes et opinions d 'experts, sur le sujet a l'ordre du jour. Le conflit de ces opinions joue un role essentiel dans !'exploration complete des apories que comporte le sujet, et dans la resolution de ces apories. II semble, par ailleurs, que toutes ces opinions puissent etre considerees comme des endoxa, par application directe de la definition meme de ce terme dans les Topiques. Des !ors, on devrait pouvoir considerer qu'il y a bien des conflits entre endoxa, et que ces conflits ont pour lieux privilegies les questions les plus importantes et les plus difficiles que le philosophe puisse se donner la tache de resoudre. Mais Aristote admet-il, dans les Topiques, que les endoxa (j'entends, naturellement, les endoxa veritables, pour reserver le cas des endoxa apparenls, sur lesquels prennent appui certains des raisonnements eristiques, 100b23 ss) puissent i\tre en conflit? Cela reste a voir. Bolton donne assez rapidement une reponse affirmative a cette importante question, en disant {p. I 97) que les endoxa comprennent a la fois les vues de la maj orite et celles des experts {cf. en effet _la definition disjonctive de 100b2! -23), et qu'Aristote sait bien que sur les problemes qui meritent une discussion, les vues de ces deux groupes tendent a etre en conflit, aussi bien d'un groupe a l'autre qu'a l'interieur de chacun des deux groupes. Pour documenter cette derniere affirmation, Bolton renvoie au chapitre I I I des Topiques. Ce chapitre, qui est consacre aux problemes dialectiques, decrit effectivement ces derniers comme des questions sur lesquelles ii y a, soit absence d 'opinions en un sens ou en l'autre (oudeleros doxazousin, !04b4), soit conflit d'opinions (enanlios, sc. doxazousin) entre la majorite et les experts, soit encore conflit d'opinions a l'interieur de chacun de ces groupes. II est done incontestable, et d' ailleurs evident, qu 'ii existe des conflits d 'opinions. Mais la question reste de savoir si ces conflits enlre opinions ( doxazousin) sont des conflits enlre endoxa. Ou, pour poser la meme question en d 'autres termes : une opinion qui est en conflit {reel et non pas seulement possible, bien entendu) avec une autre opinion, a quelque groupe d 'hommes qu'elle appartienne, peut-elle etre encore comptee comme un endoxon, ou bien est-elle inapte a i\tre comptee comme telle, du seul fait de cette relation de conflit ?

246

• 1'

' '·

11·

J , BRUNSCHWIG

Une premiere raison de penser qu'il ne peut y avoir de conflit entre endoxa est que Ies conflits entre opinions que Bolton considere comme des conflits entre endoxa sont evoques par Aristote a propos de la notion de probleme dialectique, alors que la notion d'endoxon sert essentielle- · ment a caracteriser les premisses du syllogisme dialectique. La distinc­ tion entre premisses et problemes est un element fondamental dans !'articulation conceptuelle du livre I des Topiques ; elle s'y presente de fagon recurrente, soit sous forme directe (prolasis vs. problema), soit sous forme paraphrastique (ex Mn hoi logoi vs . peri Mn hoi sullogismoi, !Olbl3-16, cf. !03b39). Cette distinction, a laquelle Bolton ne fait aucune reference, doit servir de fil conducteur pour resoudre la question qui nous occupe, celle de l'eventuel conflit des endoxa. Pour qu'un probleme, c'est-il-dire une question de forme « est-ce que p ou non-p ? » (I 4, IOI b32-34), puisse constituer un probleme dialeclique, c'est-a-dire un probleme qui « merite d'etre discute» ( I 1 1 , !05a4), il faut qu'il « contienne une difficulte» (echein aporian, !04a7, !05a9). Le conflit des opinions relatives a cette question, sous quelque forme qu'il se presente (entre la majorite et les experts, entre experts, etc.), peut servir d'indice ou de critere de la presence d'une telle difficulte. Mais le caractere « problematique )) meme du « probleme 1> emp€:che chacune des deux reponses qui peuvent lui etre donnees (p OU non-p) de pouvoir jouer le role de premisse dans un syllogisme dialectique, c'est-a-dire de compter comme un endoxon. En effet, si la premisse dialectique doit etre un endoxon (!04a8-10), c'est pour qu'il soit aussi difficile que possible au repondant de la refuser au questionneur. Dans l'hypothese OU un endoxon A pourrait etre en conflit avec un autre endoxon B, le repondant n 'aurait que trop de facilite a refuser A, que lui propose le questionneur, en prenant lui-m€:me appui sur B . C'est une premiere raison de penser que l'hypothese d'un conflit entre endoxa n'est pas envisagee par Aristote, dans la description de la methode dialectique qu'il presente dans les Topiques. Examinons cependant quelques arguments que l'on pourrait essayer d'avancer dans le sens inverse. Si Bolton a pense pouvoir �tiliser la description du probleme dialectique que contient le chapitre I 11 a l'appui de sa supposition qu'un conflit entre endoxa etait toujours possible, c'est peut-etre parce que, dans le chapitre I 4 (!Ol b28-36), Aristote presente la difference entre prolasis et problema comme une difference de pure forme. Aussi longtemps qu'elle est une question a laquelle on est plus OU moins force de repondre par !'affirmative, en raison du caractere endoxon de cette reponse, une prolasis est en effet une interrogation simple, de forme « est-ce que p ? ». Un problema, de son cote, est une interrogation double, de forme « est-ce que p ou non ? ». A ce premier niveau de son analyse, Aristote peut done prendre des exemples qui ont le meme contenu materiel pour illustrer la prolasis (« est-ce que

)

REMARQUES' SUR LA COMMUNICATION DE ROBERT BOLTON

247

' vivant terrestre bipede' est la definition de l'homme ? ») et pour illustrer le problema (« est-ce que ' vivant terrestre bipede' est la definition de l'homme, OU non ? i>) ; ii precise meme qu '« il partir de toute profasis on peut faire un problema en changeant [simplement] la forme » (melaballon loi lropoi, IOl b36). II pourrait done sembler legitime de deduire de la que des endoxa peuvent etre en conflit. En effet, si un probleme dialectique est une question sur laquelle les experts (par exemple) sont en desaccord, parce qu 'ils lui donnent des reponses contradictoires, chacune de ces reponses est un enonce sur lequel ils sont en desaccord, parce que les uns l'approuvent et que les autres le rejettent. Mais, s'il est vrai qu'il suffit d'un changement de forme pour transformer une premisse en probleme et un probleme en premisse, chacune des reponses contradictoires apportees par les experts a une question problematique paralt pouvoir jouer le role de premisse. Comme elles sont en conflit, ii doit y avoir des conflits entre premisses. Et comme les premisses dialectiques sont des endoxa, il doit y avoir des conflits entre endoxa. En realite, ce raisonnement est un non sequilur. II repose sur l'oubli de la distinction entre premisses et problemes d 'une part, premisses dialectiques et problemes dialectiques d' autre part. Ce que dit Aristote au chapitre I 4, concernant la difference purement formelle entre premisse et probleme, ne vaut que pour les premisses et les problemes en general. Mais « ii ne faut pas poser que toute premisse est dialectique, ni que tout probleme est dialectique » (I 10, I04a4-5). Outre les differences formelles qui separent la premisse dialectique en tant que premisse et le probleme dialectique en tant que probleme, ii faut compter avec les differences epistemologiques qui les separent en tant que premisse dialectique et que probleme dialectique : le probleme dialectique souleve une aporie, il engendre ou peut engendrer des conflits d'opinions ; la premisse dialectique est un endoxon. Si l'on considere un probleme dialectique, comme «le monde est-ii eternel ou non ? » (I04b8), ii ne peut etre question d'attribuer le statut d'endoxa aux enonces contradictoires (<de monde est eternel », « le monde n'est pas eternel ») dont la contradiction meme le qualifie comme probleme ; c'est au contraire le fait que chacun a ses partisans autorises qui le qualifie comme probleme dialectique. Mais, dira-t-on peut-etre, la possibilite d'un conflit entre endoxa n' est-elle pas inscrite dans la definition meme de ces endoxa ? Cette definition, on s'en souvient, est en effet disjonctive : sont endoxa <1 les choses qui sont acceptees (a) par tous, ou (b) par la plupart des gens, ou par les sophoi, et pour ceux-ci soit (c) par tous, soit (d) par la plupart, soit (e) par les plus celebres et les plus distingues » (100b2I-23, traduction inspiree de celle de Bolton, p. 189). Dans la section 8 de son etude (pp. 208-212), Robert Bolton soutient, avec beaucoup de plausibilite, que l'ordre dans lequel Aristote enumere les divers types d'endoxa n'est

I .

; i

I '

i Ii I

248

J,

BRUNSCHWIG

pas accidentel : la liste va, en ordre decroissant, des types dotes (au mains en premiere analyse) de l'autorite maximale aux types dotes (toujours en premiere analyse) de l'autorite minimale. Que l'ordre de la liste ne soit pas dil au hasard, c 'est certainement tout a fait exact. Mais ii serait inexact de croire que la question de savoir si les endoxa d'un niveau superieur d'autorite peuvent entrer en conflit avec des endoxa d'un niveau inf€:rieur d'autorite s'en trouve reglee pour autant. En effet, deux interpretations differentes de cette hierarchie peuvent etre envisagees (je laisse de cote la question de savoir s'il peut y avoir des conflits entre endoxa de meme niveau, par exemple entre opinions egalement admises par tous : il me semble qu'aucun texte ne permet de penser qu'Aristote admettrait une telle possibilite). Dans la premiere interpretation de la hierarchie des types d'endoxa, il peut y avoir des conflits entre endoxa de niveaux differents, parce que ce qu'on peut appeler le caractere endoxal, ou l'endoxalite, d'une opinion lui appartient absolument, si et seulement si elle appartient it l'un des types d'endoxa distingues par Aristote, et quels que soient par ailleurs le nombre et la qualite des gens qui ne la partagent pas. Supposons que, sur une question donnee, la plupart des hommes croient que p, et que la plupart des experts croient que non-p : p sera un endoxon parce qu'il appartient au type (b) de la definition, et non-p sera aussi un endoxon parce qu'il appartient au type (d) de la definition. Ainsi peut-on obtenir sans aucune difficulte, dans cette interpretation de la definition, deux endoxa contradictoires. Toutefois, les declarations ulterieures d'Aristote ne semblent pas all er dans ce sens : lorsqu 'il reprend la lisle des types d 'endoxa pour definir la premisse dialectique (104a8-l l), il prend soin de preciser que les opinions des experts ne comptent comme des endoxa que si elles ne sont pas des paradoxa, c'est-il-dire si elles ne sont pas en conflit aveC les endoxa de niveau superieur, ceux qui sont acceptes par tous Jes hommes ou par presque tons ; en effet, dit Aristote, on pourrait parfaitement poser (comme premisse dialectique) ! 'opinion des experts, « si (du mains] elle n'est pas contraire aux opinions de la majorite de� hommes» (!04a l l -12). L'endoxalite d'une opinion experte p depend done, non seulement de ce que pensent les experts, mais aussi de ce que pense simultanement la majorite sur la m�me question : si la majorite pense que p, p est certes un endoxon, mais pour une raison qui n'est plus la caution des experts, et qui est precisement la caution de la majorite ; si la majorite pense que non-p, c'est non-p qui est un endoxon, et p n'en est pas un, malgre la caution des experts ; si la majorite n'incline ni du cote p ni du cote non-p, p est alors un endoxon, pour la raison precise que les experts lui donnent leur caution. Je n'ai pas de desaccord sur ce point avec Robert Bolton, qui dit tout ce que je viens de dire tres clairement et tres exactement : je me

RE MARQUES -, SUR LA COMMUNICATION DE ROBERT BOLTON

249

borne it renvoyer au commentaire qu'il donne, p. 210, de la definition aristotelicienne de la premisse dialectique (104a8-13). Cependant, il croit bon d'ajouter : « Naturellement, I' opinion d'un expert continue a compter, par definition, comme un endoxon (100b21-23), meme si elle est contrai�;' it !' ? pinion de la majorite » (p. 210, souligne par moi). La, je cesse d etre d accord : car ce n'est nullemen t la definition aristotelicien­ ne de l'endoxon qui implique cette conception, laquelle a son tour . 1mphque la poss1b1hte du conflit des endoxa : c'est seulement !'interpre­ tation particuliere que donne Bolton de cette definition. En effet, la disjonctivite de la definition initiate de l'endoxon est ambigue. On peut la comprendre d'abord comme suit : est endoxon ce qui repond it la description (a), et aussi bien ce qui repond it la descr1pt10n (b), etc. Dans ce cas, l'adjonction de me paradoxos , en 104al 0- l l , est une veritable adjonction, qui modifie le sens de la definition ; mais alors les deux textes deviennent virtuellement contra­ dictoires, une opinion paradoxal e d'expert devant compter, selon 104al 0-1 1 , comme n'elant pas un endoxon, et selon 1 00b21 -23, comme etanl un endoxon. Cette difficulte disparait si l'on considere que la disjoncti� ite de a definition initiate doit s'interpreter, par anticipation, . a la Ium1ere de 1 adJonct1on me paradoxos, c 'est-8.-dire si on la comprend _ _ ams1 : est endoxon ce qui repond it la description (a), ou bien, s'il n'existe sur le sujet considere aucune opinion repondant a la description (a), ce qui repond it la description (b), etc. Dans cette seconde interpretation de . _ la hierarchie des types d 'endoxa, l'adjonction me paradoxos de 1 04a l 0- l l ne modifie pas le contenu de la definition initiale ; une opinion d'expert ne �omptera comme un endoxon que s'il n'existe pas, sur le sujet cons1dere, d'opinion determinee admise par tous les hommes, ou par p�esq_u e tous. L 'endoxon en question ne sera done en conflit, par _ defimt10n (en vertu de la definition ainsi comprise), avec aucun autre endoxon. Et de meme dans les autres cas de figure. Les difficultes dans lesquelles se trouve engage Bolton pour avoir adopte la premiere interpretation se constatent imm
!

250

I .

J, BRUNSCHWIG

REMARQUES SUR LA COMMUNICATION DE ROBERT BOLTON

discussion dialectique. L'ennui est que ce mot vague de « standing» occulte la differenciation, ici plus necessaire que jamais, entre prem1sse _ et probleme. Une opinion paradoxale, si elle repond a certa1 � es , conditions determinees (etre soutenue par un ph1losophe celebre, ou b1en pouvoir s'appuyer sur une argumentation rai sonnable), acquie �t effecti­ _ _ vement un « standing» dialectique, ma1s a_ litre de probleme dialecltque, c'est-a-dire a titre de question qu'il est legitime de prendre pour su3et d 'une discussion dialectique : Aris tote precise expressement que Jes lheseis sont une sous-espece des « problemes » (dialectiques) (104b29-34), en ctepit de !'usage laxiste « actuel» qui donne le nom de lheseis a tous Jes « problemes dialectiques » (104b34 ss). II n'est pas non plus 1mposs1ble, dans ces conditions, qu 'une thesis obtienne une autre forme de . « standing» dialectique, a savoir celle d' � ne th�se de �endu e avec succes : . . ii se peut en effet qu'a !'issue de I entretien dialectique, la lhesi_s paradoxa]e de !'expert ait resiste a un effort de refutatIOn : Cette lhe� LS . acquiert alors le statut de concluswn dialectiquement non d1squahfiee, comme nous l'avons vu ci-dessus. En revanche, non seulement le « standing» dialectique des theseis ne signifie nullement qu'elles puis�ent etre des endoxa, mais encore la definition de leurs d1verses especes (104bl9-28, 31-34) implique qu'aucune ne peut etre un endoxon. II _ s'ensuit, si j e ne me trompe, d'une part, qu' � ucu �e lhe� is ne p �u � serv1r de premisse dialectique, puisque toute prem1sse dialectique do1t etre un _ endoxon ; d'autre part, que le conflit entre une thesis et Jes opmwns qu1 la contredisent n'est pas un conflit entre endoxa. . Reste encore un argument a considerer, celu1 qu1 prendra1t appu1 sur !'introduction des degres d'endoxalite. C'est, d'une fa�on generale, un des grands merites de l'etude de Robert Bolton qu � d'avoir attire . entre la !'attention sur une difference importante, et souvent neghgee, description initiale de la methode dialectique, au debut des Topiques ( l OOalS-21), et la description recapitulative de cette meme methode vers la fin des Refutations sophistiques (34, 183a37 ss, cite par Bolton p. 199). II releve a tres juste titre que la nouvelle description introduit, � n to �t cas formellement, une nouveaute importante : le ra1sonnement dialect1que, ou peut-etre un certain type de raisonnement dialectique est ici '. decrit comme un raisonnement qui procede, non pas s1mplement a partir d'endoxa, mais plus precisement ek ton huparchontl'm hOs endoxotato � (183a38), c'est-a-dire a partir de ce qui est le plus endoxon, ou de ce qm est aussi endoxon que possible. Laissons de elite, pour le moment, la question de savoir si cette formulation nouvelle constitue une innovation substantielle dans la description de la methode dialectique, et si cette innovation signifi_e que l'objet de cette description n'est plus la dialectique en general, ma1s une forme particuliere de la dialectique, celle qu'Aristote desig�e ici �ous le nom de peirastique. Cette question fera l'ob3et de notre sectwn smvante. .

�

I

.

.

251

Demandons-nous seulement, dans le cadre de la question qui nous occupe maintenant, si l' acceptation de degres d' endoxalite implique ou non la possibilite de conflits entre endoxa. Si un enonce p est plus endoxon qu 'un autre enonce q, cela implique evidemment que p et q sont tous deux endoxa, quoique a des degres divers. Mais cette situation peut­ elle se realiser lorsque p et q sont contradictoires ou incompatibles ? Pour Bolton, la reponse affirmative ne fait pas de doute. II ecrit par exemple (p. 1 99) : « [la nouvelle formulation donnee dans Jes Refutations] suggere . . . que si les premisses d'un argument sont endoxa, et si meme elles sont tirees du sous-ensemble consistant maximal des endoxa, mais si elles sont en conflit avec d' autres endoxa qui sont plus endoxa qu'elles­ memes ne le sont, alors !'argument qui prend appui sur elles ne sera pas un argument proprement dialectique, ou du mains ii ne sera pas le meilleur argument dialectique (possible), puisqu'il ne prend pas appui sur ce qui est le plus endoxon parmi Jes endoxa pertinents sur le sujet considere ». On voit clairement dans ce passage que, selon Bolton, pour qu'un argument A soit dialectiquement disqualifie au profit d'un argument B, ii faut que Jes premisses PA de A soient en conflit avec des premisses plus endoxa P8, qui sont Jes premisses de B. Deux conditions sont done ici conjointes : (I) Jes PA sont plus endoxa que Jes P8 ; (2) Jes PA sont en conflit avec Jes P8. A mon avis, cependant, ces deux conditions, loin de pouvoir se conjoindre, s'excluent mutuellement. Si la seconde est realisee, nous l'avons deja vu, ou bien les PA ne sont pas endoxa, ou bien les P8 ne le sont pas, ou bien enfin ni les unes ni les autres ne le sont; de sorte qu 'ii n'y a pas entre elles inegalite d'endoxalite. II me reste, bien siir, la charge de montrer comment la premiere condition peut et doit etre realisee sans que la seconde le soit. On peut le faire brievement, me semble-t-il, de la fagon suivante, en adoptant un schema simplifie autant qu'il est necessaire. Entre le sujet et le predicat de la conclusion a prouver, le raisonnement consiste a intercaler un moyen terme. Supposons qu 'une meme conclusion puisse etre atteinte, dans !'argument A, par !' intercalation d'un moyen terme « savant» M'\ et, dans !'argument B, par l'intercalation d'un moyen terme « non-savant )> M8. Les premisses (( savantes » PA sont des endoxa s'il n'y a pas d'opinion com_mune ou generate, en un sens ou en l'autre, sur Jes rapports entre MA et Jes termes de la conclusion. Les premisses r non­ savantes » P" sont des endoxa si Jes rapports appropries entre M8 et Jes termes de la conclusion sont endosses par }'opinion commune ou generale. Dans ces conditions, (1) ii n'y a pas et ii ne peut y avoir de conflit entre Jes pA et Jes P8, la difference entre Jes moyens termes MA et M" rendant inaptes a se contredire Jes premisses oil figurent ces moyens termes ; (2) Jes PA et Jes P8 sont, Jes uns et Jes autres, des endoxa ; (3) Jes ps sont plus endoxa que Jes pA, parce que leur endoxalite se situe a un

. i; I •:

i i I

252

i.

, ·

i. .

'

J. BRUNSCHWIG

n1veau superieur sur rechelle des degres d'endoxalite ; par suite, !'argument B est dialectiquement preferable a !'argument A. On peut conclure de cette analyse, me semble-t-il, que l 'hypothese d'une possibilite de conflits entre endoxa ne trouve pas d'appui dans la notion des degres d'endoxalite. Cette notion souleve un autre probleme, auquel j'ai fait allusion ci­ dessus. Lorsqu'Aristote introduit, dans la description de la methode dialectique du chapitre final des Refutations, !'expression ex endoxotaton, formellement differente de !'expression ex endoxon, qu'il avait employee dans la description initiale des Topiques, entend-il faire comprendre que ce qu'il decrit maintenant n'est plus qu'une espece particuliere de ce qu'il decrivait initialement ? Telle est la question que je voudrais discuter a present.

3. Du DEBuT DES ToPIQUES

'

.I. '

A

LA FIN DES REFUTATIONS

Revenons done au texte recapitulatif des Refutations sophistiques (34, 183a37 ss, cite par Bolton p. 199), pour l'etudier a un autre point de vue que dans la section precedente. L'une des differences qui separent ce texte de la presentation initiale de la methode dans les Topiques (I 1 , lOOalS-21) est, nous l'avons vu, qu'Aristote s'ecarte (ou parait s'ecarter) de sa premiere description du raisonnement dialectique. Dans celle-ci, le raisonnement dialectique etait de_crit comme un raisonnement qui procede « a partir d 'endoxa ». Dans la nouvelle description, Aristote introduit (ou parait introduire) une nouvelle exigence, en demandant a la fois au questionneur et au repondant de se contraindre a ne plus raisonner qu'«3. partir de ce qui est le plus endoxon, ou de ce qui est a_ussi endoxon que possible» (183a8, b6). Quelle est la signification de cet ecart ? Notons au passage qu'il ne peut i\tre mis sur le compte d'une hypothetique evolution de la pensee d ' Aristote entre les Topiques (au sens etroit du terme) et les Refutations, meme s'il peut y avoir d'autres traces d'une telle evolution : car, comme le signale tres honni\tement Bolton lui-mi\me, p. 200, !'expression ek ton endoxon continue a servir a la caracterisation des arguments dialectiques dans les Refutations, au chapitre 2 (1 65b3-4). Cependant, pour Robert Bolton, cet ecart entre les deux textes, et la realite substantielle de ! 'innovation qu'il represente dans le second par rapport au premier, ont une signification fondamentale et servent d'appui essentiel pour son interpretation. I l leur associe etroitement une autre difference qui separe les deux textes, et qui concerne la designation meme de la methode decrite !!ans chacun d'eux : dans les textes du livre I des Topiques, celle-ci s'appelle simplement « la dialectique» ou « le raisonnement dialectique )) ; dans le texte recapitulatif, en 183a39, la

REMARQUES SUR LA COMMUNICATION DE ROBERT BOLTON

253

tache a laquelle il s'est agi de fournir ses moyens est decrite comme celle de « la dialectique kath'hautim et de la peirastique ». Bolton traduit cette expression de la fagon suivante : « la dialectique au sens strict et la peirastique» (p. 199 et p. 200) ; je reviendrai plus loin sur cette traduction. En couplant les deux innovations qui caracterisent le second passage, par rapport au premier, on obtient ainsi l'hypothese suivante : la regle qui impose seulement de raisonner a partir de ce qui n'est qu'endoxon caracteriserait de fagon minimale la dialectique en tant que telle, sous toutes ses formes y compris les moins exigeantes, tandis que la regle qui impose de raisonner a partir de ce qui est le plus endoxon (( ne serait pas exigible dans certaines des formes ou usages de la dialectique, et le serait seulement dans d 'autres, en particulier dans la forme particuliere de la dialectique qui s'appelle la peirastique» (p. 200). Avant d 'apprecier cet element capital de !'interpretation de Robert Bolton, je presenterai d 'abord une observation generale, qui peut avoir quelque interi\t sur le plan de la methode. lei comme dans un cas precedemment etudie, nous nous trouvons en presence de ce qu'on pourrait appeler des reiterations a variantes dans le texte d'Aristote. Tout a l'heure, deux passages distincts ( I 1 , 100b21-23 et I 10, 104a8l l), l'un precedant l 'autre de quelques pages, decrivaient les divers types d 'endoxa ; le second contenait une precision absente du premier {3. savoir, que les opinions d'experts sont endoxa a condition de ne pas i\tre en contradiction avec les opinions de la majorite). Dans le cas qui nous occupe maintenant, on observe une relation analogue entre deux textes, cette fois beaucoup plus eloignes l'un de l'autre. Le second introduit une determination differente, et formellement plus stricte, du caractere des premisses dialectiques {ex endoxotaton vs. ex endoxon) ; en outre, il semble fournir les moyens de rendre compte de cette difference en laissant supposer, moyennant !'interpretation appropriee d'une expression diffi­ cile, que l'objet decrit n'est pas exactement le meme dans les deux descriptions (la dialectique en general vs. une ou plusieurs formes particulieres de la dialectique). Le probleme d'exegese se pose en termes voisins dans les deux cas : lorsque, par rapport a une description anterieure A, une description posterieure B introduit des elements formellement nouveaux, faut-il en conclure que l'objet decrit par B n'est pas, ou a de bonnes chances de n'i\tre pas, le meme objet que l'objet decrit par A, OU bien doit-on penser que l'objet des deux descriptions reste le meme, les elements formelle­ ment nouveaux apportes par la description B devant i\tre lus, en quelque sorte par anticipation, comme sous-entendus dans la description A ? Nous avons vu plus haut (pp. 248-9) que dans le cas de la description des divers types d'endoxa, la seconde solution paraissait devoir etre preferee a la premiere, pace Bolton. I l pourrait se faire qu'il en soit de meme dans le cas present, c'est-a-dire que le requisit d'endoxalite maximale,

' '

:I !I : '1 ' .

' ·'

"Il'i

'•I;i

255

J. BRUNSCHWIG

REMARQUES SUR LA COMMUNICATION DE ROBERT BOLTON

formellement exprime dans le second texte seulement, ne soit qu'une fa�on de reecrire plus precisement, apres tout le chemin parcouru entre temps, le requisit d'endoxalite tout court, tel qu'il figurait dans le premier texte. La plus grande simplicite de !'expression, dans ce premier texte, s'expliquerait facilement : a son niveau, qui est celui de la toute premiere phrase du traite, la notion d'endoxon n'a pas encore ete definie ; a plus forte raison ne pouvait-il etre question d'y faire usage de la notion de degres d'endoxalite. Correlativement, dans cette ligne d 'interpreta­ tion, on essaiera de reduire I 'impression que la description iniiiale et la description finale puissent ne pas decrire le meme objet, et !'on s'efforcera de donner de ! 'expression litigieuse « la dialectique kath'hauten et la peirastique » une interpretation qui preserve l'identite de l 'objet decrit, a savoir la dialectique tout court. C'est cette ligne d'interpreta­ tion que je voudrais esquisser maintenant. II y a d 'abord une raison generale de penser que le texte recapitulatif des Refutations n'introduit pas substantiellement de requisit supplementaire par rapport au texte programmatique du debut des Topiques, et, par consequent, que l'innovation formelle qu'il presente n'est pas destinee a caracteriser une forme particuliere de la metbode dialectique par opposition a d'autres : c'est precisement que le second texte se presente comme l'exacte recapitulation de l'accomplissement d'un certain programme, qui n'est autre que le programme decrit par le premier (simplement augmente du programme propre des Refutations, dont Aristote fait discretement remarquer, en !83bl3-15, qu'il ne figurait pas dans le programme initial des Topiques). Cette recapitulation se presente explicitement comme telle (1 83a34-36 : « II nous reste, apres avoir rappele noire projet initial, a en dire brievement quelques mots, et a mettre le point final a notre discours »). Elle se deroule conformement a l'annonce qui la precede immediatement : nous nous proposions,�
presenter sa seconde description comme celle de l 'accomplissement exact de la tache decrite par la premiere. Un point de discussion plus particulier, sur le meme sujet, doit concerner, comme je l'ai dit plus haut, !'interpretation de la formule « la dialectique kath'haute.n et la peirastique » (183a39-bl), qui designe ce dont la tache est decrite dans le passage des Refutations qui nous interesse. Bolton traduit, comme nous l'avoris vu : « l a dialectique au sens strict et la peirastique », et la suite de ses considerations sur la peirastique semble montrer qu'il comprend cette formule dans le sens de « la dialectique au sens strict, et notamment la peirastique », peut-etre meme dans le sens de « la dialectique au sens strict, c'est-A-dire la peirastique ». En tout etat de cause, cette traduction peu orthodoxe de kath'hauten aurait besoin d'etre plus precisement argumentee et defendue, en particulier contre des possibilites concurrentes, comme « la dialectique consideree en elle-meme» (c'est-a-dire : abstraction faite des differences qui pourraient separer ses diverses formes OU especes). Cette possibilite peut elle-meme s'accommoder, au mains provisoirement, de diverses hypotheses concernant le statut de la peirastique par rapport a la dialectique « en elle-meme ». On peut en effet comprendre, soit « la dialectique en elle-meme, et aussi la peirastique », celle-ci etant alors consideree comme exterieure a la dialectique, comme elle l'est dans la presentation du chapitre 2 des Refutations (1 65a38-39, b3-7) ; soit « la dialectique en elle-meme, et notamment la peirastique », celle-ci etant alors consideree comme une espece ou une partie de la dialectique, comme elle !'est dans le chapitre 8 (169b25) et dans le chapitre 1 1 (! 7l b4-6). Dans ces diverses hypotheses, entre lesquelles j e ne me prononcerai point ici (car je suis moins certain que ne !'est Robert Bolton que !'on puisse reunir en un tableau coherent toutes Jes indications que donne Aristote sur le statut de la peirastique), ii est en tout cas clair que l'emploi de !'expression de « dialectique kath'hauten » peut avoir ete motive par le souci de souligner que l 'objet de la description reste precisement la dialectique « en elle-meme», c'est-3.-dire abstraction faite de la diversite de ses formes, c'est-3.-dire encore sous loules ses formes ; peu importe, a cet egard, que la peirastique mentionnee a cote d'elle soit l'une de ces formes OU qu'elle ne le soit pas. Cette interpretation de la formule litigieuse permet done de preserver un rapport de stricte adequation entre le programme initial des Topiques et le bilan recapitulatif des Refutations : l'objet decrit ne s'est pas modifie restrictivement d'une description a l'autre. La distinction dont nous venons de discuter Jes bases textuelles et conceptuelles opposait, dans ! 'interpretation de Robert Bolton, une dialectique ex endoxon et une dialectique a exigences plus fortes, ex endoxotaton. Bolton superpose encore a cette opposition celle qui separe une dialectique qui ne servirait qu'a s'exercer l'esprit et une dialectique

254

256

4. DIALECTIQUE GYMNASTIQUE, DIALECTIQUE NON-GYMNASTIQUE

'I

I

,

I i

' I

I : ! '

'

BRUNSCHWIG

qui aurait vocation d'intervenir serieusement dans la recherche de la verite et dans la j ustification de la connaissance. Suivons maintenant Jes formes que prend notre debat a ce nouveau plan.

I, I

I

J.

I

I

L

Aristote, on le sait, distingue plusieurs « usages » ou « utilites i> de la dialectique, ou plus exactement de son traite de dialectique (pragmaleia, 10l a26), dans le chapitre I 2 des Topiques. La premiere de ces r ntilites» est celle d'un entrainement ou d'une gymnastique (gumnasia) ; parmi Jes « utilites i> non-gymnastiques figurent diverses formes d'« utilites » scien­ tifiques et/ou philosophiques. Dans la logique de son interpretation, Robert Bolton est tout naturellement conduit a faire coincider !'opposition (< gymnastique vs. non-gymnastique)> avec !'opposition « dialectique ex endoxOn vs. dialecti­ que ex endoxolalon ». Si la regle qui impose de raisonner a partir de ce qui est le plus endoxon etait veritablement une regle plus contraignante que celle qui impose de raisonner simplement a partir de ce qui est endoxon tout court, ii serait effectivement tentant de considerer que la regle la moins contraignante est celle qui gouverne !'usage purement gymnasti­ que de la dialectique, alors que la regle la plus contraignante est celle qui gouverne Jes usages philosophiquement et scientifiquement interessants de la methode dialectique, la « mise a l'epreuve» et la « recherche » (peira, skepsis, 159a33). Comme le dit Bolton en resumant sa these sur ce point, « la dialectique a visee gymnastique a d'autres regles que la dialectique destinee a la ' mise a l'epreuve'. et a la ' recherche ' . . . et la dialectique gymnastique n'a pas toujours necessairement besoin de raisonner a partir de ce qui est ' plus endoxon ' » (p. 201). Remarquons d 'abord qu'il semble, au moins a premiere vue, aSsez paradoxal de supposer que l'on puisse s'entrainer a une pratique quelconque en observant des regles differentes de celles qui gouvernent cette pratique. On peut cependant admettre que dans certains cas, l'on puisse s'exercer a pratiquer telle OU telle activite en observant des regJes moins strides que celles qu'il faudra respecter lorsqu'on se livrera a cette activite « pour de bon », ou encore en n'observant qu'une partie de ces regles : on s'entraine au tennis, par exemple, en « faisant des balles », sans respecter toutes Jes regles qui gouvernent le jeu pendant une veritable partie de tennis. Mais ce modele n'est pas le seul concevable en la matiere, et l'on trouverait sans peine d'autres activites a propos desquelles ii serait passablement absurde de pretendre s'entrainer a Jes exercer sans en observer integralement les regles. En ce qui concerne l'exercice dialectique, la question peut done se poser de savoir lequel de ces deux modeles est pertinent. Je ne pense pas

REMARQUES SUR LA COMMUNICATION DE ROBERT BOLTON

257

que l'on puisse durcir !'opposition entre « dialectique gymnastique» et « dialectique non-gymnastique», comme le fait Bolton, au point d'assi­ gner a chacune d'elles des regles differentes, et de Jes constituer comme deux especes distinctes, ou deux groupes distincts d'especes, de la d1alect1que. Je ne pense meme pas qu'il soit tout a fait legitime de parler de <• dialectique gymnastique » et de « dialectique(s) non-gymnastique(s) ». Et, so1t d1t en passant, je crois n'avoir suggere nulle part qu'aux yeux d'Aristote, la methode decrite dans Jes Topiques n'ait ete appropriee qu'a !'usage gymnastique et scolaire, malgre le reproche que m'adresse Bolton, p. 213. II me semble en tout cas que la methode dialectique a Iaquelle on s'entraine dans l'usage « gymnastique)> n'est pas une autre dialectique, ou une autre forme de dialectique, ou une dialectique gouvernee par d'autres regles, que celle que l'on pratique, selon Jes Topiques eux­ memes, a ? e� fins « serieuses » et non-gymnastiques. Dans le chapitre I 2, _ Anstote d1stmgue, non pas divers types ou especes de dialectiques, mais d1verses fonctions, usages OU utilites d'une methode qui est toujourS decrite au singulier, clans ses diverses relations avec une multiplicite de fins ou d 'objectifs. En ce qui concerne plus particulierement l'utilite du traite des Topiques pour la gymnastique dialectique, Aristote dit meme que cette utilite est « de soi evidente » (ex auton kalaphanes), parce qu'« une fois en possession d'une methode, nous pourrons plus facilement argumenter sur le sujet qui se presente » (10l a29-30) ; Jes autres r ntilites » du traite semblent, par contraste1 avoir besoin d'explications. Le traite permet done d 'acquerir la technique meme dont ces explications montreront l'utilite, non une technique plus faible ou moins exigeante. Et s'il en est ainsi, l'activite a laquelle se livre celui qui s'entraine a la gymnastique en question n'est pas une autre activite que « la dialecti­ que » (10l b2) elle-meme, ni une activite gouvernee par des regles d1fferentes ou moins strictes que dans ses autres applications ou usages. S'il en Hait autrement, Aristote devrait demontrer separement (1) que son tra1te permet de s'entrainer a la methode dialectique, et (2) que cet entrainement est lui-meme utile, en depit de ses regles moins strictes, a l'emploi de cette methode dans Jes tiiches r nobles» de la science et de la philosophie. C'est pourtant ce dont ii n'apparait nulle part ressentir le besoin. Bolton presente cependant a l'appui de sa these, p. 200, quelques arguments precis qu'il vaut sans doute la peine de discuter. II met en contraste Jes deux textes ou ensembles de textes suivants : d'une part, le d1ffic1le passage de V I I I 3, ou Aristote dit que pour celui qui s'entralne (gumnazomenoi), par opposition a celui qui apprend (manlhanonli), ii n'est pas necessaire de respecter la regle qui impose de proceder a partir de premisses « plus intelligibles » (gnorimoleron) que la conclusion ; d 'autre part, Jes chapitres V I I I 5-6, ou Aristote demande au contraire de

,

,'

!j

!.

•.i

258

l i

1'

J. BRUNSCHWIG

respecter une regle qui impose de proceder a partir de ce qui est « plus endoxon » et « plus intelligible» que la . conclusion. Pour eviter d'attribuer une contradiction a Aristote, on devrait supposer que, dans les deux textes, Aristote oppose diverses formes ou especes de la dialectique : dans le premier, la regle que la dialectique gymnastique peut se dispenser de respecter s'impose, au contraire, a une autre forme particuliere de la dialectique, a savoir « une certaine sorte de discussion dialectique qui peut servir d'apprentissage (scientifique)» (note 21) ; dans le second groupe de textes, symetriquement, la regle du « plus endoxon » ne s'impose qu'a une forme particuliere de la dialectique, celle qui vise « il l'epreuve et a la recherche» (peiras kai skepseils charin, 159a33), et ii est tacitement entendu que la dialectique gymnastique n'a pas a s'en preoccuper. La question est cependant de savoir si ces textes ou groupes de textes sont veritablement opposables. On peut d'abord observer que le chapitre V I I I 3 appartient a un ensemble de chapitres destines a indiquer au queslionneur Jes regles qu 'ii doit observer pour formuler et mettre en ordre ses questions (cf. V I I I I , 155b3, 18, et la clause de terminaison en V I I I 4, 159al5-16), alors qu'a partir du debut du chapitre 4, Aristote passe aux directives destinees au repondanl (159al6 ss). Quel que soit le sens exact de la difficile fin du chapitre V I I I 3 , i i est done exclu, m e semble-t-il, que l'on puisse l'utiliser pour dire qu'Aristote impose au repondanl tantot une regle plus !ache, tantot une regle plus stricte, contradiction qui ne pourrait se resoudre qu'en supposant qu 'ii a en vue, ici et la, di verses formes ou especes de dialectique qui n'obeissent pas aux memes regles. Voyons maintenant plus precisement si les textes cites imposent l'idee d'une division de la dialectique en especes distinctes. Dans le premier (3, 159al 1-14), Aristote oppose l'entralnement gymnastique (tili gumnazomenoi) a l'apprentissage (tili manthanonli) . Pour expliquer comment ii comprend ce contraste, Bolton procede en deux etapes. Dans un premier temps (note 17), ii dit que la question de savoir si la procedure destinee a l'apprentissage est ou non consideree ici qomme une espece de procedure dialectique merite discussion, mais que cette question peut etre provisoirement laissee de cote, car de toute fagon, le texte decrit clairement ce qui n'est pas requis dans la dialectique gymnastique, a savoir le respect de la regle dia gniJrimiJteriJn. Dans un second temps (note 21), ii revient sur le probleme qu'il avait provisoire­ ment laisse ouvert, et propose de le resoudre en disant qu'Aristote a ici en vue « une certaine espece de discussion dialectique qui peut etre utilisee pour l'apprentissage ». Concernant le premier point, ii me semble que Bolton commet une sorte de petition de principe : en parlant de « dialectique gymnastique », ii presuppose ce qui est precisement a demontrer, a savoir qu'il existe

REMARQUES SUR LA COMMUNICATION DE ROBERT BOLTON

259

Une « dialectique gymnastique » qui s'oppose a d'autres « dialectiques» OU especes de « dialectiques ». II n'est pas du tout indifferent meme provisoirement, que « l'apprentissage» du manthanOn soit ici c� ngu ou non comme une espece de procedure dialectique : car s'il ne l 'est pas, l'opposition entre « apprentissage » et « entrainement» met f:ln contraste une procedure non-dialectique d'enseignement et une procedure dialecli­ que d'entrainement, et l'on doit conclure que la notion d'<(entrainement» ou de <( gymnastique » suffit ici a caracteriser la dialectique comme telle, dans le contexte de son opposition avec les procedures didactiques. Que la « gymnastique» puisse passer ici pour l'embleme de la dialectique elle­ meme, cela ne me parait nullement impossible, si l'on veut bien considerer (1) que !'opposition « apprentissage vs. entrainement» se superpose, clans le contexte immediat, a !'opposition « enseignement vs. interrogation » (eriltilnli le kai didaskonli, 159a l 3), laquelle renvoie elle­ meme a une opposition classique entre enseignement scientifique­ demonstratif et interrogation dialectique (SE 2, 165b 1-3 ; 1 1 , I 72a 15 ss ; APr I I , 24a22 ss ; cf. aussi Top. V I I I I , 155b!0-16) ; (2) qu'il arrive a Aristote d 'utiliser la notion de « gymnastique J> pour determiner une caracteristique qui appartient selon lui a la methode des Topiques comme telle (cl. par exemple I 1 1 , 105a9 : un probleme dont la solution serait trop longue ne doit pas faire l'objet d'un examen, car ii _ « plus de difficulte qu'il ne convient a un exercice comportera1t gymnastique») ; (3) enfin, que la regle dia gnilrimilteron, telle qu'elle est ici im� osee a la procedure d'enseignement-apprentissage, n'associe pas a la not10n de « plus intelligible» la notion de « plus endoxon » ; ii y a done de bonnes chances que le « plus intelligible » qu 'Aristote a ici en vue (comme en 155bl0-16) soit le « plus intelligible en soi », et non pas le « plus intel­ ligible pour nous», lequel (comme le montre fort bien Bolton, pp. 207208) est le seul des deux qui puisse etre associe positivement a I' endoxon dialectique (comme ii !'est dans des passages comme 1 59b8-9 et 13-15). S'il en est ainsi, on comprend pourquoi je resiste a l'argumen­ . tat10n de Bolton dans son second temps (note 21). Nulle part Aristote ne
' "

'i

I I

260

'I

J.

REMARQUES SUR LA COMMUNICATION DE ROBERT BOLTON

BRUNSCHWIG

compte les differences entre les divers usages de la dialectique. On ne peut done pas assigner ces procedures a une dialectique didactique qui s'opposerait, a l 'interieur du genre dialectique, a la dialectique gymnastique ; ii me semble arbitraire d'invoquer sur ce point le passage 159a 1 1 , comme le fait Bolton dans sa note 21 ; et la tentative que contient cette meme note pour differencier cette didactique intra­ dialectique de la didactique extra-dialectique decrite par Aristote dans des textes comme le chapitre 2 des Refutations m'apparait comme un expedient desespere. On ne peut pas non plus soutenir, avec des textes a l'appui, que ces procedures soient inappropriees quand on pratique la dialectique a des fins purement gymnastiques, et qu'elles n'ont leur place que dans une forme (< serieuse )) et epistemologiquement respectable de dialectique, celle qui vise a la peira et/ou a la skepsis. Le chapitre V I I I 5, oil Aristote introduit ses instructions a !'usage du repondant, permet en effet de verifier avec precision que son objectif est de marquer les differences qui separent la dialectique, prise en tant que telle et dans son unite, et Jes activites voisines dont ii importe de la distinguer, a savoir le dialogue didactique et le dialogue agonistique. Revendiquant sur ce point particulier un role de pionnier (159a36-37), Aristote indique que la tache du repondant dialecticien n'a jamais ete bien definie, contrairement a celle du repondant dans un dialogue didactique (oil ii doit toujours repondre conformement a ce qu'il pense, 1 59a29) et a celle du repondant dans un dialogue agonistique (oil ii doit toujours s'efforcer de ne paraitre subir aucun dommage, 1 59a32). Entre ces deux extremes de la bonne foi- sans limites et de la mauvaise foi systematique, Aristote veut definir, pour la premiere fois, les regles d 'apres lesquelles le repondant, dans un entretien proprement dialecti­ que (tous « usages» confondus), doit accepter ou rejeter les premisses qui Jui sont proposees par le questionneur. , Je cours ici le risque, a vrai dire, que l'on me fasse une objection : en effet, les fins propres de la dialectique, dans ces premieres lignes du chapitre V I I I 5, sont designees successivement par des expressions variees : gumnasias kai peiras heneka (1 59a25), skepseos charin (28), peiras kai skepseos (sc. charin) (33). Pourquoi ces variations, dira-t-on, si elles ne sont pas destinees a differencier par leurs fins diverses especes de l'activite dialectique ? J e crois cependant qu'on peut y voir di verses fagons de se referer, par des descriptions qui mettent alternativement en jeu telle ou telle de ses finalites, une seule et meme activite, la dialectique en tant que telle. Les echos partiels que ces expressions se renvoient Ies unes aux autres en sont deja l'indice. Si, par exemple, Aristote voulait opposer radicalement !'usage gymnastique (gumnasia) et !'usage « examinateur» (skepsis) de la dialectique, ii serait singuliere­ ment maladroit de sa part de brouiller la nettete de cette opposition en associant, a quelques lignes d'intervalle, chacun de ces deux termes a un

261

meme troisieme1 a savoir peira. Le sens de ces variations est a chercher, en realite, clans.des considerations de contexte. En 1 59a25, le terme de gumnasia figure dans la description des fins de la dialectique, tres vraisemblablement, parce que c'est ce terme qui implique !'opposition la plus nette entre la dialectique et le dialogue didactique, et qu'a cet 1 endroit precis, c'est cette distinction qu il importe de mettre en lumiere (26-27, cl. 161 a25) ; en 1 59a28 et 33, c'est au contraire le terme de skepsis qui sert, seul ou en combinaison, a determiner I ' essence de la dialectique, parce que c'est ce terme qui implique !'opposition la plus nette entre la dialectique et le dialogue agonistique ou polemique, et qu'a cet endroit precis c'est cette autre distinction qui est thematisee (159a27 et 33). La structure generale du paragraphe et de la longue parenthese explicative qui en occupe la plus grande partie (159a26-26) confirmerait encore, a mon sens1 cette analyse. Cette structure se presente en effet ainsi : (I) Les regles a ! 'usage des repondants sont indeterminees, pour ceux qui argumentent en vue de la gumnasia et de la peira. (2) En effet, les buts ne sont pas les memes pour ceux qui enseignent ou apprennent et pour ceux qui polemiquent, et ils ne sont pas non plus les memes pour ceux-ci (c'est-a-dire !'ensemble de ceux dont ii vient d 'etre question) et pour ceux qui discutent entre eux a des fins de skepsis. (3) En effet, d'une part, celui qui apprend sait quelle regle ii doit suivre ; d'autre part, ceux qui polemiquent le savent aussi ; mais dans les rencontres dialeciiques, pour ceux qui argumentent non en vue de I' agOn, mais en vue de la peira et de la skepsis, on n'a pas encore articule clairement les objectifs que doit viser le repondant, ni les regles qu'il doit suivre dans ses reponses. (4) Puisqu'il n'existe pas de tradition sur ce point, essayons done d'en dire nous-memes quelque chose. On le voit clairement, je crois : le raisonnement n'offre une trajectoire satisfaisante que si l'on suppose qu'a travers les variations de son expression, Aristote garde ici en vue une activite uni.q ue1 l'activite dialectique en tant que telle, qu'il s'agit de differencier par rapport a ses voisines extra-dialectiques, la didactique et l'agonistique ; les descrip­ tions sous lesquelles ii est fait reference a la dialectique varient en fonction des differentes delimitations qu'il s'agit de souligner successive­ ment. On rendrait incomprehensible, en revanche, le deroulement du discours aristotelicien dans ce passage si l'on croyait qu'a chaque description differenciee de la dialectique correspond un objet differencie, a savoir une sous-espece differenciee de cette meme dialectique. '

i.

I

'

I

262

J. BRUNSCHWIG

BIOLOGIE. l.OGIQUE Er Mtf}.Pf[YSIOUE

seminaire

chez AlusrorE 1987

CNRS-N.S.F.,

Editions du CNRS, Paris, 1990.

5. EN GUISE DE CONCLUSION PROVISOIRE II est temps de conclure cette serie de remarques, qui n'ont certainement pas atteint le fond du debat. Celui-ci se situerait, je pense, au niveau des rapports entre l'utilite gymnastique de la methode des Topiques et son utilite philosophique et scientifique. La tendance generale de !'interpretation de Robert Bolton est de marginaliser, au moins relativement, !'usage gymnastique de la methode des Topiques, et, du meme coup, de neutraliser tous les aspects de la dialectique aristotelicienne qui lui sont lies (codification des roles dialectiques du questionneur et du repondant, institutionnalisation scolaire du dialogue, postulat de la non-contradiction des endoxa) ; c'est seulement a ce prix, selon lui, que !'on peut comprendre que la methode puisse avoir une utilite qui ne soit pas seulement gymnastique, et que !'on puisse passer sans heurt de son usage philosophique, tel qu'il est decrit dans les To piques, a la pratique de cet usage philosophique dans les traites. J e pense, pour ma part, que !'usage philosophique de la dialectique, selon !' auteur des Topiques, est solidaire de son usage gymnastique, plus qu'il ne !'est de !'usage philosophique (futur) qui en sera fait dans les traites. Mais ii serait trop long de s'expliquer la-dessus, a cause en particulier du soupgon de couches redactionnelles distinctes que !'on peut avoir par rapport au chapitre I 2 (cf. eli de en ·101a36). J'aurai siirement d'autres occasions de poursuivre avec Robert Bolton une discussion dont j ' ai deja tire grand profit. Pour finir sur quelque chose de plus beau et de plus significatif que mes laborieux commentaires, je me contenterai de rappeler (meme si je ne remplis pas toutes les conditions pour le suiv're) le celebre conseil que donne Parmenide a Socrate dans le Parmimide platonicien (!35d), conseil que tous les exegetes, depuis Alexandre d'Aphrodise, ont rapproche non sans raison des Topiques aristoteliciens : « exerce-toi, entraine-toi dans cette activite qui a l'air d'etre .inutile et que la plupart des gens appellent du bavardage, tant que tu es encore jeune ; sinon, la verite t'echappera ».

' ;1

i'

'1 1I ll

ii

COMMENTS ON ROBERT BOLTON'S THE EPISTEMOLOGICAL BASIS OF ARISTO TELIAN DIALECTIC

DANIEL DEVEREUX

Aristotle mentions in a fragment that Zeno was the discoverer of dialectic! It was Plato, however, who gave the activity its name, using the term dialeklike to refer to the special method of inquiry characteris­ tic of the philosopher. In the Republic dialectic is tendentiously linked to the theory of separate, paradigmatic forms: to be a dialectician, or philosopher for that matter, is to be an adherent of Platonic forms:' fo later dialogues . we find a less partisan view. The dialecbcian is concerned with unchanging forms, but there is no requirement that these forms serve as paradigms or exist separately from their instances.' Plato may still have been an adherent of Platonic forms, but professing allegiance to this doctrine is no longer a condition of being a dialectician or philosopher. The metaphysical commitments are relaxed to the point where there is room for a good deal of disagreement among dialecticians regarding the nature and status of forms; even such young upstarts in the Academy as Aristotle may qualify as dialecticians. What remains constant through Plato's dialogues is the (I) Aristotle, Soph., fr. I Ross. In the Metaphysics (987b29-33, 1078bl7-30) Aristotle implies that dialectic originated with Socrates and Plato, but in these passages he seems to be using the term in its specific Platonic sense, designating a systematic method for the discovery of the essences of things. In the Parmenides (135c-136c), Plato attributes to Zeno a method of discussion that is at least a close cousin of his dialectic; the Zenonian type of discussion is recommended as an exercise that will help to prepare the young Socrates for the attempt to define such notions as the "beautiful", the "just" and the "good". (2) See, e.g., 532a-b, 533c-d, 510a-5l l e. (3) Parmeni�es 135b-c, Sophist 253c-254a, Philebus 57e-58a.

264

' , ,

D. DEVEREUX

THE EPISTEMOLOGICAL BASIS OF ARISTOTELIAN DIALECTIC

exalted position of dialectic in relation to other disciplines. Its objects are primary in the sense of being most intelligible and prior to the obj ects of other sciences or disciplines. In the late dialogues as well as in the Republic, to be a philosopher is to be an accomplished dialectician. In spite of the fact that Aristotle devoted an entire treatise to the subject of dialectic, its role or function in his philosophy is not at all clear. The array of views on this question held by recent commentators is nothing short of bewildering. Some argue that Aristotle retains the Platonic conception of dialectic as the method of philosophicar inquiry; not only is it the preferred method in dealing with subjects that do not admit of scientific demonstration, e.g. ethics and metaphysics, but even in the sciences it provides the only possible justification for their first principles and undergirds our confidence in scientific procedures and methods.' At the other extreme are those who contend that Aristote­ lian dialectic is a rule-governed method of discussion whose aim is simply to increase one's ability to argue effectively on both sides of a question.5 According to this view, dialectic has no concern for getting at the truth and has no role in a properly philosophical or scientific inquiry. One naturally wonders how these two groups can be talking about the same thing; moreover, if the second view is correct, one wonders why Aristotle would use the term "dialectic" for an activity that seems to have so little in common with its Platonic namesake.•

It seems to me that these two groups of commentators are not talking about the same thing. Certainly the texts they regard as crucial for their views are not the same and, as I will try to indicate later, it is not at all clear that both sets of texts are concerned with what Aristotle calls "dialectic". Between the two extreme views we have mentioned there are of course a number of intermediate positions that have been staked out and argued for by scholars, both recent and not so recent. Robert Bolton's view is an example of one of these more moderate positions. He argues that the second extreme view is correct in regard to the kind of dialectic Aristotle is mostly concerned with in the Topics. The sort of organized discussion or debate that is in the forefront of the discussion in this work is what might be called "gymnastic" dialectic. The point of this kind of discussion is to train the participants to argue effectively - whatever the topic - on the basis simply of opm10n or what seems plausible . The participants in such discussions are not supposed to be concerned with the truth of either the premises or the conclusions of their arguments. In view of these characteristics, one can easily understand why gymnastic dialectic would have no role in philosophical or scientific inquiry. However, according to Bolton, there is another kind of dialectic, called "peirastic", which is discussed most clearly in the Sophistical Refutations, but also alluded to at various points in the Topics. In peirastic dialectic one is concerned with truth; Bolton argues that it is a requirement that the premises of peirastic arguments be true. And since this form of dialectic relies on what is commonly known to be true (la koina), it will be useful for the testing of scientific or philosophical hypotheses. So the weakness of the second extreme view - the view that dialectic has no role in philosophical or scientific inquiry - is that it focuses exclusively on one form of dialectic, "gymnastic" dialectic, and ignores peirastic dialectic, a form well suited for testing theories or hypotheses. But peirastic provides only one way of testing theories; an equally (if not more) important way is by appeal to data derived from perceptual observation. When first discovered, such data will not

(4) M. Nussbaum, Fragility of Goodness, 240-58. (5) See, for example, P. Moraux, "La joute dialectique d'apres le huitiBme livre des Topiques", in G . E . L. Owen, ed., Arislolle on Dialeclic (Oxford, 1968), 288-9; also J. Brunschwig, "Aristotle on Arguments Without Winners or Losers", Wissenschaflskolleg Jahrbuch 1984/5 (Berlin, 1986), 40. Moraux and Brunschwig are of course aware that in Topics I 2 dialectic is said to be useful in relation to the "philosophical sciences" (see Moraux, 3 1 1 , Brunschwig, 34). Their view seems to be that whatever the philosophical utility of dialectic might be, it is not discussed in the Topics and does not seem to have influenced the description of the method of conducting dialectical discussions in Book V I I I . Although they d o not say so, i t seems likely that Moraux and Brunschwig would regard the claim about the philosophical utility of dialectic as problematic; given their view that dialectical discussions have no concern for truth as such. (6) A number of commentators believe that Aristotle composed the Topics during the period he was a member of the Academy. If so, the contrast between Aristotle's and Plato's conceptions of dialectic is an interesting indication of the range of disagreement tolerated in the Academy at this time. If we suppose that the procedures described in Topics V I I I have some resemblance to actual practices in the Academy (notice that 101a26-30 seems to presuppose that "dialectical exercises" already take place; Aristotle's treatise will be useful to participants in these exercices), it is difficult to believe that Plato would have regarded such discussions as examples of dialectical inquiry; perhaps he would have considered them a useful propaedeutic to dialectic (cf. Parmenides 135c-I36c), but no more than that. Aristotle came to believe, on the other hand, that proficiency in such discussions was the mark of a dialectician. Moraux suggests that within the Topics we can see evidence of an evolution from a Platonic to an Aristotelian conception of

265

dialectic. The main body of the work, Books II-VII, reflects a Platonic view of dialectic as a method of inquiry aimed at the discovery of essences and definitions; it is only in the later sections (Books I, V I I I) that we find the contrast between philosophy and dialectic and the related contrast between arguments kala doxan and kal' atelheian. On the chronological stratification of the Topics and Sophislical Refutations, see the interesting and generally plausible suggestions of H. Maier, !Jie Syllogislik des Arisloleles, I I 2 (Tiibingen, 1900), 78, n. 3 (78-82); also J . Brunschwig, Arisiole: Topiques, I-IV (Paris, 1967), LX 1-LXXXI I I, and J. Barnes, "Proof and Syllogism", in E. Berti, ed., Aristotle on Science: The Poslerior Analylics (Padua, 1981), 43, n. 43. For further remarks on chronology, see below, n. 19.

.

·:. 1 266

D. DEVEREUX

qualify either as endoxa or as things commonly kno�n to be true , and _ thus will not fall within the scope of pe1rast1c testmg. Dialectw, or more exactly peirastic dialectic, is therefore not the only, or even the _ most fundamental, method of philosophical or scientific inquiry, as the first extreme view would have it. Before beginning our test of this hypothesis by examining the relevant texts in the Topics and Sophistical Refutations, I should perhaps indicate some general points of agreement and disagreemellt with Bolton's interesting and valuable discussion . I am not persuaded that Aristotle recognizes in these works a special form of dialectic intended for use in philosophical inquiries. As will become clear shortly, my understanding of the nature and function of peirastic is quite different from Bolton's. On the other hand, I believe Aristotle regarded dialectic as useful to philosophy, even if it is not a method to be used in philosophy. Its philosophical utility is only partly a matter of . improving one's mental acuity; the results of dialectical d1scuss10ns may also be useful, as we shall see. Although the differences that will emerge between my view and Bolton's are substantial, I beli �ve our overall positions are fairly close, at least when compared with the extreme views held by recent commentators.

1, i.

Bolton argues that there are _ indications in Topics V I I I that Aristotle distinguishes between two forms of dialectic, one used for mental training or "gymnastics", the other for "testing and inquiry" (200). In the case of the former, the premises must be plausible (endoxa)' but they need not be more plausible than the conclusion and

seems to me at least a defensible transla.tion of ta en�o�a in the the adjective endoxos as . a mo.d1fier of op1n1on.s . or chiefly because, the or1g1nal premises). Barnes prefers "reputable views" as a translation, meaning of the adjective endo:i:os is "of good repute" (see "Arist�tle and the �etho?s �� Ethics" Revue Internalionale de Philosophie [1980], 498-500). It is true that , plaus1ble Jacks th� etymological connection, but it is not clear that A.ristot.le has thi� connection in mind. Consider, for example, the passage in I 10 where dialectical premises (or endo:r:a) are characterized as views held by all or by most or by the wise-either by all of the wise or by most or by the most notable, provided these are not contrary to gen�ral. opinion (me paradoxos): "for a man would grant (lheie gar an Lis) a view held by the \v1se 1f it does not . the conflict with generally held opinions" (104a8-12). Thus the reason fo.r .1�c.lud1ng (non paradoxical) opinions of th.e wise among ta endoxa is the fact th� t an 1n�1v1dual could _ "1n ac.cordance be expected to grant such opinions in a discussion (cf. 104a33-35: op1n1ons held with the arts" count as endoxa or dialectical premises " for a man would grant views by those who have rr1ade a study of these things"). The op�rative noti�n seems to b.e _ or what it what an interlocutor could reasonably be expected to accept in a d1scuss1on, would be unreasonable or churlish for an interlocutor to refuse to accept without offering a (7) "Plausible views" Topics ("plausible" for

THE EPISTEMOLOGICAL BASIS OF ARISTOTELIAN DIALECTIC

267

they need not be true; in the case of the latter, the premises must be both true and more plausible than the conclusion - they must indeed be "as plausible as possible", i.e. there should not be other more plausible premises from which the conclusion follows. Bolton bases his claim that in gymnastic dialectic the premises need not be more plausible than the conclusion on the following passage in V I I I 3 : One must be aware then, whenever a thesis is hard to argue against, that one or other of the aforesaid things has happened to it. Whenever, on the other hand, it is a harder task to argue for the claim or proposition [you need] than against the thesis, a doubt may arise whether such claims should be granted or not; for if a man is going to refuse to grant it and demand that it be argued for, he will be prescribing a harder task than was originally proposed; if, on the other hand, he grants it, he will be convinced on the strength of what is less convincing. If, then, one ought not to increase the difficulty of the [original] problem, it should be granted; if, on the other hand, one ought to deduce through premises that are more familiar (gnOrimOlerOn), it should not be granted. In other words, it should not be granted by one who is learning (lOi manlhanonli) unless it is more familiar; but it should be granted by one who is training {lOi gumnazomenOi) as long as it appears true. Clearly, then, the circumstan­ ces under which admissions should be claimed are different for a questioner and a teacher. (159a3-14).

supporting argument. "Reasonable expectation" is determined at least in part by social context; Aristotle is assuming a discussion between two individuals before an audience, and the social setting imposes certain constraints on the behavior of the interlocutors (Endoxa "may be true, of course, ... but it is not their truth, as such, which is of interest to the dialectician, but rather their social undeniability"; Brunschwig, "Arguments", 33; cf. 160b2-6, 1 58a7-13, SE 169b30-34, 175b33-36; cf. Moraux, op. cil., 277-8). The term "plausible" is constantly used in the context of informal argument, and its use corresponds closely to the idea of what it would he reasonable to grant to an interlocutor in a certain social setting of discussion. (It is also worth noting that Barnes translates the opposite of endoxos-adoxos-as "implausible" in the Revised Oxford Translation: see, e.g., VIII 5.) It might be argued, on the other side, that Aristotle's endoxa include more than just the views one could be expected to grant in a discussion, for in his initial characterization o( endoxa (100b21-23) he includes views held by the wise without qualification, hut, as we have seen, he holds that only non-paradoxical views of the wise could be expected to be granted in discussion. One might also point out that at 104a8-12 Aristotle is specifying the class of "dialectical premises", not the class of endoxa, so this passage s.hould not be taken as an indication of what is meant by the term endoxa. According to this view, the premises of dialectical arguments form a subclass of ta endoxa, and we should take 104a812 as a correction of the suggestion at 100a29-30 that the class of dialectical premises coincides with the class of endoxa. I-lowever, 104a8 ff. is usually taken as a fuller specification of the class of endoxa (see Brunschwig, Topiques, 1 13-4; Barnes, "Methods", 501}; if so, we must understand it as a correction of the suggestion at 100b21-23 that opinions of the wise in general-whether or not they are paraxodica[-count as endoxa. (It is usually thought that Socrates' denial of acrasia is treated as one of the endoxa, even though it clearly conflicts with general opinion amphisbetei lois phainorne­ nois enargOs; Nie. Eihics 1 145b25-28; cf. Top. 105a35-bl , and Brunschwig, Topiques, 130, n. 5. I lowever, Socrates' view is mentioned after the endoxa or phainomena have been �et forth [ 1 145b8-20, esp. line 20], and within the second stage of the inquiry, i.e. the development of the aporiai [cf. 1 1 45h2-4 with b20-22 and 1 1 46b6-8].) -

268

l

I

I

Ii

I ' i

I I,

!

269

D. DEVEREUX

THE EPISTEMOLOGICAL BASIS OF ARISTOTELIAN DIALECTIC

Thus in the case of dialectical discussions for the sake of training, i.e. what Bolton calls "gymnastic dialectic", the premises need not be more familiar or more plausible• than the conclusion. By contrast, in a discussion between teacher and student, the premises must be more familiar. Now it would be wrong to suggest that discussion between teachers and students is a type of dialectic with stricter standards than gymnastic dialectic.' In V I I I 5, Aristotle distinguishes between dialec­ tical discussions and discussions among those engaged in teaching or learning. The latter type of discussion is not a type of dialectic at all. How do we know, then, that there is a type of dialectic, different from gymnastic, which requires that premises be more familiar or more plausible than the conclusion ? Bolton points out that in V I I I 5 and 6 Aristotle says that an answerer in a dialectical discussion should only grant to the questioner what is more plausible and more familiar than the conclusion to be established. ...the premises s�cured by the questioner should all be plausible, �nd more plausible than tHe proposed conclusion, if the less familiar is going to be inferred through bhe more familiar. (!59b!3-15). Those who try to deduce from premises more implausible than the conclusion clearly do not deduce correctly; hence these should not be granted to questioners. (160al4-16) When we compare these passages with the passage in V I I I 3, we seem to be faced with two options: either (i) Aristotle is here contradicting what he said earlier, or (ii) he is thinking of a different form of dialectic ----,- one with more stringent rules than gymnastic dialectic.10 Bolton di course recommends the second option. If Aristotle does recognize a second form of dialectic which requires that premises always be more plausible than the conclusion argued for, it is odd that he does not mention it in the passage in chapter 3; ; he ought to be contrasting gymnastic dialectic not only with discussion .for the sake of teaching and learning but also with this stricter form of dialectic, for they both require that premises be more familiar than conclusions. I believe that in all of these passages Aristotle is speaking of one particular type of discussion which he calls "dialectical" . It

should be noted that he does not use an expression corresponding to "gymnastic dialectic": nowhere in the Topics do we find the noun dialeklike modified by an adjective indicating a particular type or use of dialectic. When he contrasts dialectical with other types of discussion in V I I I 5, he refers to the former as "discussions for the sake of training and examination (gumnasias kai peiras)" , or "discussions for the sake of examination and inquiry (peiras kai skepseos)" , and there is no explicit indication that these descriptions pick out different types of dialectic.11 When Aristotle stipulates in chapters 5 and 6 that the answerer should grant to the questioner only premises that are more plausible and more familiar than the conclusion, there are no signs in the context that he means this to apply to a particular type of dialectic rather than to dialectical discussions in general. He seems to be setting up a general rule for all dialectical discussions. If we look again at the problematic passage in chapter 3, it is clear, I think, that Aristotle is not implying that there is a type of dialectic which allows premises to be less plausible than the conclusion. He is envisioning a special situation in which, as he puts it, "it is a harder task to argue for the claim or proposition [you need] than against the thesis" . The difficulty he mentions is a difficulty for the questioner; the task of the questioner is to refute the answerer's thesis by establishing its contradictory. When he says that it is sometimes a harder task to argue for the claim or proposition than against the thesis, what he seems to mean is that the questioner may find it harder to establish one of the premises he needs for his conclusion than it would be to establish the conclusion once. that premise is granted. And it is important to note that this difficulty for the questioner is not one that was foreseen at the outset of the discussion - "for if [the answerer] refuses to grant [the premise] and demands that it be argued for, he will be prescribing a harder task than was originally proposed". Is it fair to impose this extra burden on the questioner? Aristotle's decision is that, while it would be appropriate in a teaching situation to demand an argument for the questionable premise, it would not be appropriate in a discussion "for the sake of training"; the premise should be granted "as long as it appears true" .12

(8) As Bolton shows (205-208), what is more familiar or more intelligible (gn6ri­ m6teron) to us (but not simply) corresponds to what is more endoxon;.see 159b6-15. (9) In n. 2 1 , Bolton suggests that there is a kind of dialectical discussion used in teaching and learning, though it is different from the learning procedure described in SE 2, 165b1-3. But it should be noted that in Top. V I I I 5, 159a26-32, discussion between teacher and student is conlrasled with dialectical and eristical discussions, and other references to didactic discussion are consistent with such a contrast; cf. 159a l l-14, 161 a2433; Rhei. 1355a21-29. (10) It is clear from the introduction to these passages (159a-32b1) that Aristotle is setting out guidelines for dialectical, not didactic, discussions.

(11) See H. Maier, op. cit., I I , 2, 67, n. 1 . (12) Aristotle here shows a concern that the tasks facing the questioner and answerer should be roughly equal in difficulty. Another example of this concern may be seen in V I I I 5. Suppose the thesis to be defended by the answerer is very plausible; this means that the contradictory of that thesis, which the questioner must try to establish, will be implausible and thus the questioner's task would seem much more difficult than the answerer's. Aristotle stipulates that in such cases the answerer should grant to .the questioner even premises that are implausible, as long as they are less implausible than the conclusion the questioner is trying to establish (159b16-19). In discussions of this kind, it

I

270

1_:!

D. DEVEREUX

THE EPISTEMOLOGICAL BASIS OF ARISTOTELIAN DIALECTIC

I think it is clear from the context that Aristotle regards this as an exceptional situation. In the normal case, the tasks of the questioner and answerer are more equally balanced, and the questioner will be able to find premises that are either more plausible than the conclus10n or at least easier to argue for. Aristotle is not telling us that whene ver we are engaged in a dialectical discussion for the sake of trammg we need .not observe the rule that premises must be more plausible than conclus10n; he is saying that although in general our premises should be more plausible or more familiar, in this. exceptional situat10n they need not . be. It is also worth noting that m the situat10n envis10ned, m which one of the premises granted is less plausible than the conclusion, it is not the case that there is another more plausible premise that can do the job; the questioner still achieves the goal of arguing from premises that are "as plausible as possible" in the given case -" I conclude that the problematic passage in chapter 3 provides no clear support for the view that there are two forms of dialectic. As far as I can see, we search the Topics in vain for a distinction betweens forms or types of dialectic. .

i

II

I

' I' "

'I II,: r;

: Ji

1. I I!

Perhaps Bolton would agree with this much: the evidence in the Topics for a special type of dialectic adapted for use in scientific and philosophical inquiry is rather skimpy and of dubious value, at least i � . . comparison with what we find in the. Sophistical Refulalwns (SophislLCL Elenchi: SE) . In several places in this work Aristotle discusses a type of inquiry or argument which he calls peirastike, and at one point he refers . to it as a "part" (meros) of dialectic -" Accordmg to Bolton, peirastic arguments are distinguished from ordinary dialectical arguments by two main features: (i) their premises must be not just plausible (endoxa), as in ordinary dialectic, but "most plausible" (endoxalala) ; (ii) their premises are things known by people in general, and hence are true, whereas the premises of ordinary dialectical arguments need not be true." I f this is right, it seems plausible to suppose that peirastic is is desirable that the answerer's thesis should be neither very easy nor very difficult to derend {cf. 105a7-9 with 158a31-bl5). Although, as Brunschwig points out, dialectical . discussions are "arguments without winners or losers", they nevertheless contain features characteristic of what we might call a fair contest. It seems reasonable to suppose that Aristotle builds in these features in order to make dialectical discussions more like a game that is interesting and enjoyable to its participants. (13) Thus it seems to me that Bolton is wrong in suggesting (at 201) that V I I I 3, 159a10-14 waives the requirement that premises be "as endoxon as possible". (14) 169b25; cf. 1 7 lb4-6. (15) The claim that the premises of peirastic arguments must be "most plausible" {endoxalala) is based on SE 183a37-bl ("We set out to discover a faculty of reasoning

·

271

more than a mere instrument for exercising the mind. If peirastic arguments proceed from things known, their primary concern must be truth, not opinion. Surely arguments of this sort would have a role in scientific or philosophical inquiry. There is an immediate problem with Bolton's characterization of peirastic arguments which raises doubts about his interpretation. The combination of the two distinctive features of peirastic premises implies a correlation between plausibility and truth such that any member of the set of "most plausible" views must be true. But it seems possible for a view to be as plausible (or endoxon) as one likes and still be false. More to the point, since our immediate concern is the interpretation of Aristotle, we might wonder whether he held such a view. Bolton apparently thinks he did not, for he says at one point that for Aristotle a premise may be "most plausible" and at the same time false -" Are we to say, then, that Aristotle's characterization of peirastic arguments is inconsistent with his view of the relationship between plausibility and truth? Clearly we need to examine carefully the evidence offered by Bolton in support of his interpretation of peirastic. Aristotle's conception of peirastic is interesting in its own right, quite apart from the question of its role, if any, in philosophical inquiry. What I propose to do is to comment on the three or four passages in the SE in which peirastic is discussed, pointing out along the way my differences with Bolton. A couple of the passages are notoriously difficult, and I am not confident that my interpretations are much more than stabs in the dark."

about any given question on the basis of the most plausible premises possible; for this is the function (ergon) of dialectic proper (dialeklikes kalh' hauten) and of peirastic."); see Bolton, 199, 234. The characterization or dialectical and peirastic arguments at SE 2, 165b3-6 might be taken to imply that the premises of peirastic arguments need not be endoxa. Bolton argues against this, appealing both to 183a37-bl and to the fact that peirastic is described as a part or type of dialectic. The claim that the premises of peirastic arguments must be true is based on Aristotle's statement that such arguments rely on koina, i.e. things commonly known; if the premises of .Peirastic arguments are things known by all, they must be true and must also be believed (this explains why it is said at 165b2 that peirastic premises are views held by the answerer); see 48-50. E. Weil also holds that peirastic is a special form of dialectic designed for use in scientific or philosophical investigations, though he does not try to support his view in detail in the way that Bolton does; see "La place de la logique dans la pensee aristotelicienne", in Revue de Metaphysique el de Morale (1951), 283-315, at 302 (English translation in Barnes, Schofield, Sorabji, eds., Articles on Arislolle, vol. I [London, 1975], 88-112, at 99). (16) See pp. 32, 64. (17) Moraux believed that Aristotle's comments about peirastic in the SE do not add up to a coherent view of its nature and relationship to dialectic; see op. cit., 388, n. 3; cf. Brunschwig, "Argu �ents", 35.

I11

:!

I

273

D. DEVEREUX

THE EPISTEMOLOGICAL BASIS OF ARISTOTELIAN DIALECTIC

In chapter 2 of the SE Aristotle distinguishes four types of arguments conducted through dialogue (en toi dialegeslhai), peirastic being one of them. There are four types of arguments used in discussions between two interlocutors: didactic, dialectical, peirastic and contentious (eristikoi). Didactic arguments are those that deduce from the principles appropriate to each subject and not from the opinions held by the answerer (for the learner must take on trust); dialectical arguments are those that deduce from plausible premises (endoxOn), to the contradictory of a given thesis; peirastic arguments are those deduced from premises held by the answerer and which anyone who claims to possess the knowledge is bound to know (in what manner is explained elsewhere); contentious (or eristic) arguments are those that deduce or appear to deduce a conclusion from premises that appear to be plausible but are not so. (165a38-b8). The parenthetical comment, "in what manner is explained elsewhere", would seem to indicate that a fuller explanation of peirastic arguments was given in another text which unfortunately has not survived.18 This would explain the rather cryptic character of the account we are given. The premises of peirastic and didactic arguments are in one respect opposites: the former are, while the latter are not, opinions or views held by the answerer.19 Dialectical premises may or may not be

believed by the answerer; what is essential is that they are plausible. Peirastic premises may or may not be plausible · what is essential is that they are believed by the answerer. Wh is this essential? And what does Aristotle mean when he says that peirastic arguments are based on things that "anyone who claims to . possess the knowledge 1s bound to know"? Several passages in later chapters give us at least a few clues.

272

(18) A note at the bottom of the Oxford translation (retained in the Barnes­ Lawrence revised version) refers the reader to Topics V I I I 5, but nothing in that chapter helps to explain the characterization of peirastic premises given here. At 165b9-10 Aristotle mentions that dialectical and peiraStic arguments are "discussed elsewhere" (en allois). Dialectical arguments are of course discussed in the Topics, but the treatise as it has come down to us contains no discussion of the kind of argument here referred to as peirastic. Perhaps when Aristotle wrote this passage, he planned to revise the Topics, so as to include a discussion of peirastic arguments but he never got around to it--or perhaps the later, revised version was lost (e.g., noting the parallel between SE 2 and Topics: I 1 , one could easily imagine the insertion o f a discussion o f peirastic arguments i n : the latter). If Aristotle is referring to the Topics when he says that dialectical arguments have been discussed en allois, it is hard to square this with the suggestion made by many commentators that the SE should be regarded as the last book of the Topics. The final chapter of the SE, especially 183a27-bl6, indicates an intention to unify this \'(Ork with the Topics, but the two treatises as we have them do not seem to represent the fulfillment of that intention; see Brunschwig's judicious remarks in Topiques, XIX-XX. (19) "Didactic arguments" reason from the principles appropriate to each subject, and not from lhe opinions held by lhe answerer" (165bl-2); why does Aristotle add the second, negative clause? (He does not do this in the case of any of the other three types of arguments.) He must think that a special warning is appropriate here: some might believe that didactic (didaskalikoi) arguments reason from opinions held by the student, but this is a mistake. In Topics V I I I 5 Aristotle himself seems to subscribe to the view here rejected: "the learner should always grant [only] what seems to him to be the case" (159a28-29; cf. 1 59a4-14}. In Topics V I I I , the activity of teaching takes the form of a question and answer discussion. The teacher asks questions-and the premises he "requests" must always be true {1 59a29-30, 161a24-28}-while the student answers in accordance with his beliefs. In the SE, on the other hand, teaching does not proceed through questioning: "Thus the one who makes this claim seerns to be unaware of the

�

F.or peirastic (he peirasli/ce) is a type of dialectic (dialekli/ce) and has in view not the man who knows but the one pretends to know but does not.20 (171 b4-6). This passage suggests that the function of peirastic is to examine those who make false claims to knowledge, presumably with the aim of _ showmg that they do not know what they claim to know.21 This helps us to see why the premises of peirastic arguments must be views held by the answerer. The answerer is assumed to be someone who claims to know some subject. I f the examination of this person's claim is to be effective, he must answer in accordance whith his beliefs; for if he does not, the results will be inconclusive (the answerer can say that he was merely "playing along" with the questioner, and that his claim to know was not really undermined.22 So in order for peirastic to fulfill its difference between teaching (didaskein) and dialectical discussion (dialegeslhai}, and that the teacher should not ask questions but make things clear himself while the other should ask questions" {171a38-b2; cf. 171 b3-4, 1 72a15-21; also Plato, Sophisf 217c-d). Thus there are two interesting differences between the Topics and the SE in regard to teaching: (i) the Topics regards tea�hing as proceeding through questioning while the SE does not; (ii) according to the Topics, the premises in a didactic argument must accord with beliefs of the answerer while this is explicitly denied in the SE. (Moraux distinguishes between "discussion qui a en vue !'instruction" and "discussion philosophique", and suggests that the passage in Topics V I I I 5 concerns the former while the passage in SE 2 concerns the latter; �P· cil., 288, n. 2. liowever, the same terms are used in each case (didaskalia, . dtdasketn}, and as far as I can see there is no basis for the distinction suggested by Moraux.} Barnes argues that the two different views of didactic argument represent two stages in Aristotle's thought regarding dialectic and its relationship to science, and that the earlier view is the one that sees teaching as involving a question and answer discussion · see "Aristotle's Theory of Demonstration", in Barnes el al., eds., Articles on Arisfofle, vol. i (London, 1975), 8 1 . If this is correct, SE 2 should be regarded as later than Topics V I I I , and w� have an additional bit o f evidence i n favor o f the chronological scheme suggested by Maier (see above, n. 6). The passage concerning teaching and learning in V I I I 5 is presumably one that would have been revised if Aristotle had carried out his intention of unifying the Topics and the SE; see preceding note. (20) In SE 2, peirastic arguments are treated, not as a subspecies of dialectical arguments, but as a different kind of argument. How, then, can peirastic be a type or species of dialectic? We will come back to this problem in a m Oment; see below, n. 34. (21} See Maier, op. cil., II 1, 67, n. 1 (on p. 68); also Moraux, op. cif., 288, n. 3, and Brunschwig, "Arguments", 35. (22) We may recall that this tactic is employed by Callicles in the Gorgias (499b-c); it is not in the end an effective facesaving maneuver since the questioner can simply point out that someone who really knows his subject would have no reason to avoid giving sincere answers.

·r1[I 274

.\

'�

:i ! '

I!

I I

I

[, ', 11 I I I. :

D. DEVEREUX

function of exammmg those who make false claims to knowledge, it must base its arguments on views held by the answerer. Peirastic examination is a special type of examination of false knowledge claims. If a geometer uses his knowledge of geometry to show that someone who claims to know geometry is a fraud, this would not count for Aristotle as a peirastic examination. As the following passage in chapter 1 1 indicates, a peirastic examination is one in which the questioner does nol rely on the sort of expert knowledge that the answerer claims to possess. Dialectic is also peirastic; for peirastic (he peiraslikt) is not the same so�t of . t have even if he accomplishment as geometry, but one that a P.ers.on m1g � were not in possession of knowledge. For it 1s poss�ble ev�n for �ne without knowledge of a subject to conduct an exam1nat1on (peiran labein) of one who does not know, if granted points [premises?] taken not from things he knows23 or from �he proper r.rinciples b1.1;t from �he consequ�nc�s which a man may know without knowing the art 1n question, but which 1f he does not know he is necessarily ignorant of the art. So it is clear t�at peirastic (he peiraslike) does not consist in knowledge of any definite subject. (172a21-28). Although parts of this passage are obscure, the main points Aristotle wants to make are sufficiently clear. First of all, he is maintaining that peirastic enables a questioner to unmask someone who pretends to he an expert even though the questioner does not possess such expert knowledge himself." Secondly, he is suggesting that the key to the ability to do this lies in those things r nne could know w1tho�t knowing the art in question, but which if he does not know he is necessarily ignorant of the art". In other words, if the art in question is, e.g., geometry, the peirastic examiner will need to know some matters pertaining to geometry; if he knows nothing at all about geometry, he is _ not going to be able to show up someone as a fraud. I f the quest10per knows enough about geometry to be able to show that the order man fails to know something that any geometer would know, he will be able ' (23) Or possibly: "from the things from which he knows" (i.e. the principles of the science; ex hOn oiden). In the immediately following lines (25-27), it is said that the peirastic examiner's argument will rest on "consequences which a man may know without knowing the art in question". Presumably the man conducting the examination will know these consequences if he is using them in his argument, and thus the premises of his argument will be drawn "from things he knows". So if ex h6n oidin is understood as "from things he knows'', what Aristotle says here seems to be contradicted in lines 25-27, whereas if we understand the phrase to mean "from the things from which he knows" there will be no contradiction, for these things-i.e. the principles-are of course distinct from the consequences. (24) This naturally reminds us of Socrates. In fact, virtually everything that Aristotle says about peirastic in the SE can be seen as reflections on Socratic practices and claims; as Brunschwig suggests ("Arguments", 35) Aristotle seems to be attempting to locate his own conception of dialectic in relation to Socrates' practice.

THE EPISTEMOLOGICAL BASIS OF ARISTOTELIAN DIALECTIC

275

to accomplish his peirastic mission. According to this view, the peirastic examiner must know more about the subject, or at least more about a part of the subject, than the person he is attempting to unmask.25 The kinds of things he needs to know, and to use in his arguments, are not the principles of the art or science but the things that follow from these-the "consequences" (hepomena). These passages in chapter 1 1 give us a clearer idea of the meaning of the original characterization of peirastic arguments in chapter 2. Once we see that the aim of peirastic arguments is to examine false claims to knowledge, it becomes clear why such arguments are based on views held by the answerer, and on things "such that anyone who claims to possess the knowledge" is bound to know them. III It is not a coincidence that the kind of examination which Aristotle calls "peirastike" is discussed within the framework of an analysis of sophistical arguments. After all, the most striking examples of pretenders to knowledge are the sophists. According to SE I , the sophist's aim is to appear wise without actually being so. The sophist holds that it is better to appear wise without being so than to be wise without appearing so, and he claims that his special art guarantees the appearance of wisdom; even those who are wise and knowledgeable may not for instance, appear so, if, they are in a confrontation with a sophist. The appearance of wisdom is achieved through skill in verbal combat. The sophist takes on all comers in debate; he argues circles around his opponents, making them seem weak and defenseless. His favorite weapon is refutation. The sophist is not afraid of combat with experts in various fields, for his art enables him to get the better of them on their own ground. If, then, a refutation is a sort of deduction, an argument depending on an accident will not he a refutation. It is, however, just in this that the experts and in general those who possess knowledge (hoi epistemones) are (25) According to his statements in the Apology, Socrates was able to show through questioning that others did not know what they claimed to know (see, e.g., 23a}. His superiority over others, his "wisdom'', consisted in his greater awareness of his ignorance, not in a superior knowledge of the subjects his interlocutors claimed to know. Thus Socrates seems to be committed to the claim, in the Apology at least, that it is possible for a questioner to unmask someone's false claim to knowledge even though the questioner does not know any more about the subject than the person he is questioning. In the Charmides Plato considers a claim like this and argues against it (170a- 17lc); it is clear from the passage we are considering in the SE that Aristotle also rejects the idea that one could disprove someone else's false claim to know even if one knew no more about the subject than the p;etender.

I

, , ! :·

i i

. ·1

.I

Ii

1

I

i,

.

�{i II'

';r. I

I

I

I

276

D. DEVEREUX

refuted by those without knowledge; for the latter use deductions depending on accidents in arguing with the knowledgeable, and these people who possess knowledge, because they are unable to draw distinctions, either say "Yes" to their questions or are thought to have said "Yes" even though they have not.26 (168b4-10). If a sophist is successful in a confrontation with an expert, he appears to show that his interlocutor does not know what he claims to know (and does in fact know). One of the sophist's ploys is thus very similar to what the peirastic examiner does: the latter actually does show that his interlocutor does not know what he claims to know. Aristotle notes the close connection between sophistry and peirastic examination in the following passage. By a sophistical refutation or deduction I mean not only a refutation or deduction that only appears true [i.e. valid], but also a refutation or deduction which is true but only appears to be appropriate to the subject. These are arguments that refute not in accordance with the subject, showing [the answerer] to be ignorant, which was the function of peirastic. Now peirastic is a part (meros) of dialectic, and this may deduce a false conclusion through the ignorance of the answerer. But sophistical refutations, even though they deduce the contradictory of the answerer's thesis, do not make clear if he is ignorant; for even those who know are tripped up by these arguments. (169b20-29). The expressions "in accordance with the subject" (kala to pragma) or "in accordance with the art" are quasi-technical in the SE and mean "in accordance with the appropriate principles of the art or science".27 We might call deductions or refutations fitting this description genuinely " scientific". According to the first part of the passage, not all sophistical refutations are fallacious; some are perfectly valid refutations, but are nevertheless sophistical in that they put up a false appearance of being scientific. So sophistical refutations are of two kinds, fallacious and (valid but) pseudo-scientific. It is pretty clear in this passage that Aristotle is envisioning · an encounter between a sophist and someone with knowledge ("for even those who know are tripped up by these [sophistical] arguments"); in such cases the sophist will try to refute his interlocutor with' a pseudo­ scientific argument, thus making it appear that he is wiser and that the other man does not know what he claims to know. Aristotle points out that showing the ignorance of those who make a false pretence of knowledge is the proper business of peirastic. A peirastic argument "may deduce a false conclusion through the ignorance of the ·

{26) It is clear that "those without knowledge" who refute the knowledgeable are the sophists, for one of the standard ploys of the sophist is deduction depending on accidents; see SE 5, 166b28-36. (27) See esp. 170a31-34; also 1 7 l b l l-12, 16-22.

THE EPISTEMOLOGICAL BASIS OF ARISTOTELIAN DIALECTIC

277

answerer". In other words, the false conclusion of the peirastic argument is reached only because of the ignorance of the answerer, and thus makes clear that ignorance. The sophistical refutation, on the other hand, reaches its conclusion ("the contradictory of the answerer's thesis") not through any ignorance of the subject matter on the part of the answerer but through a deduction which is inappropriate to the subject matter.28 A peirastic argument is thus the counterpart of a particular kind of a sophistical refutation-the kind designed to make it appear that someone who has knowledge is ignorant. Peirastic arguments are designed to show the ignorance of someone who merely pretends to have knowledge. Since the sophist is one who aims at the appearance but not the reality of knowledge, peirastic is a technique to be used first and foremost against the sophist. Thus it seems appropriate that peirastic examination should be discussed within the framework of an analysis of sophistical arguments. Aristotle believes that a skillful dialectician should be able to hold his own in a confrontation with a sophist. A practitioner of dialectic must have the same skill and knowledge that the sophist has.29 Of course, he will not use fallacious arguments in a dialectical discussion. But in order to avoid such arguments, and in order to be able to detect unintentional fallacies in the arguments of his interlo­ cutor, he must understand all of the various ways in which arguments can be fallacious. For without dialectical skill we can fall prey in our own arguments to the very same fallacies that the sophist knowingly perpetrates.30 A skillful dialectician will be able to hold his own against a sophist both as answerer and as questioner.31 I f the sophist attempts to refute a thesis maintained by the dialectician, the dialectician will be able to block any sophistical deduction by pointing out the fallacy involved. (He will be like the subject who ruins a magic show by explaining to the audience how the magician's trick is performed .32) On the other hand if the sophist challenges one to question him and attempt {28) Others besides sophists are guilty of offering arguments which, though valid, are inappropriate to the subject matter. A favorite example is Bryson's method of squaring the circle; cf. SE 1 7 l b l 2-18, 172a2-7, Post. Analylics 75b37-76a3. Plato is sometimes charged with this failing; cf. Gen. and Corr. 316a 1 1-15. {29) See SE 170b8- 1 1 , 172b5-8. In the Rhel. Aristotle says that sophistry and dialectic are the same power or capacity: 1355b15-21 ; cf. Melaph. 1004b17-26, SE 165a2431. (30) See, e.g., SE 175a5-12. (31) Aristotle advises dialecticians to avoid confrontations with sophists; cf. Top. 164b8-12 with 159a30-32. Presumably this is why he does not mention the ability to deal with sophists as one of the ways in which dialectic is useful. (32) So�rates has this deflationary role in the Eulhydemus. ,

1 1,

·

.

11I'

.

:1

"

I

.

:1:1 1 . I

I

'I

:\ . I I

1 1' :

: ' '. �

i

' '

I , i

D. DEVEREUX

to show that he does not know something,33 the dialectician will be able to do just that. He will do it in the way we have seen, using the technique of peirastic examination. . . . Just as an "extra-mural" discussion in which peirastic exam1nat1on is used is quite different from an intra-mural dialectical discussion, so also peirastic arguments are quite different from th ose used in dialectical . discussions. But peirastic is not an art or skill in its own right; it is rather a particular application or use of the skill of dialectic,"" This view of the relationship between peirastic and dialectic seems to be borne out in the following passage from SE 1 1 .

So then clearly peirastic is not a knowledge or science of any definite subject; and thus it is concerned with everything, for all .the ar�s make use of certain common things as well. Hence everybody, 1nclud1ng laymen, makes use in some way of dialectic and peirastic; for all to some extent undertake to test those who profess to know things. What serves t�em here is the common things (la koina),35 for they know these themselves JUSt

,; l , , , I

'

278

!

·I

(33) See, e.g., Hipp. Majo' 287a-b, , . (34) We are now in a position to suggest a solution to the puzzle mentioned earlier (see n. 20 above): How can peirastic be a type or species of dialectic if peirastic arguments are not a type or species of dialectical arguments? Consider the relationship between sophistry and dialectic. If it is true that these two names refer to the same skill or capacity {see above, n. 29), then when a sophist skillfully formulates sophistical arguments, he is using the skill or techne of dialectic. Of course, sophistical arguments a �e _ Is not a type or species of dialectical arguments (see Top. I00a29-10 l a4, SE 165b�-8); I� rather that the skill of dialectic is such that it can be used to formulate not only dialectical . arguments, but also other, very different� sorts of argum :nts. Si�ila:I)'. , peirastic . . arguments are quite different from dialectical, but the same skill, d1al �ct1c, ts 1nv�lved in . the discovery and formulation of both. Moraux also seems to view pe1rast1c as a particular way of using dialectic, and he suggests that Aristotle has peirastic in mind when he speaks of the usefulness of dialectic "for purposes of casual encounters" (pros de tas enteuxeis; 101a30-34); op. cif., 290, n. 3. Certainly one kind of "extra-mural" encounter a dialectician might have is with a false claimant to knowledge, in which case a peirastic examination might be the appropriate response. But Aristotle does not seem to have . anything quite so specific in mind in I 2, and he speaks not of refuting hut of persuadingof bringing the interlocutor over to our view (see the parallel use of met�bibazein in 161a2936; cf. Rhel. 1355a24-39, and Eud. Ethics 1216h26-30). Bolton (214) infers from the �act . that peirastic is a "part" of dialectic that peirastic arguments must he d1alec�1cal . arguments, and hence must "employ endoxa as premises". It �hould be clear, in the h�ht of what has been said above, why I think this inference is questionable; the natural reading of SE 2 conflicts with the view that peirastic premises are a type of dialectical premises, and we are not compelled to view them as such by Aristotle's reference to peirastic as a part of dialectic. (35) "What serves them here is the koina" translates laula d' esli la koina. As Brandis suggests, something like hois chrOnlai hoi idiolai must be supplied after laula. Bolton (215) translates II. 31-32 as: "For everyone attempts to test those who profess knowledge on a limited basis, and this basis is the common things". The usual translation of mechri linos is "to some extent", or something similar (cf. Rhel. 1354a3-6): it is not clear to me that the Greek words can bear the sense Bolton gives them (i.e. "on some basis"). Bolton's translation is slightly tendentious in that it could be taken to suggest that the premises of peirastic arguments consist of the koina. The passage implies that peirastic arguments rely on the koina, but it does not necessarily imply that the koina are the premises of such 0.rguments.

THE EPISTEMOLOGICAL BASIS OF ARISTOTELIAN DIALECTIC

279

as well as the scientist, even if what they say seems wide of the mark. Thus they all practice refutation; for they engage unskillfully in the activity that dialectic performs skillfully, and he who examines by means of the art of deduction is the dialectician. (172a27-36). When we "undertake to test" or examine those who profess to have knowledge, we make use of dialectic; and the person who examines­ who is peirastikos-by means of the art of deduction is the dialectician. Peirastic is evidently being viewed as a particular use or application of the skill of dialectic. It is characteristic of dialectical arguments in general, and of peirastic in particular, that they do not involve any scientific knowledge but rely instead on certain "common things". What are these elusive "common things"? I believe Bolton is right in maintaining that they cannot be universal principles of demonstration." Are the koina then, as Bolton maintains, things that everyone not only believes but knows to be true? Let us first note that reliance on koina is not a special feature of peirastic; in several passages in the SE we are told that it is characteristic of the dialectician to rely on koina, and the art of rhetoric apparently shares this feature with dialectic.37 I f to rely on koina means to make use of premises commonly known to be true, the implication (clearly unacceptable) of these passages would be that all dialectical and rhetorical arguments are based on true premises. A closer look at Aristotle's use of the term will make it clear that it does not mean "things commonly known to be true". Aristotle sometimes uses koina to designate generally held opinions. Consider, for example, the following passage in Rhel. I I . Moreover, in relation to some [interlocutors], even i f we have the most exact knowledge, it will not be easy to persuade on the basis of that knowledge. For argument based on knowledge implies instruction, and this is impossible (i.e. in relation to some people); it is necessary to make use of the koina in framing our arguments and producing conviction, as we pointed out in the Topics in connection with everyday encounters with the general public. (1355a24-29). The passage referred to in the Topics mentions that if we become familiar with generally held opinions, we will be in a better position to persuade the man in the street by appealing to his own convictions (36) Pp. 215-216. As Bolton points out, the reference to Zeno's argument concerning motion as "common" in relation to medical questions at SE 1 1 , 172a8-9 is a clear indication that the "common things" in this context at least include more than the universal principles of demonstration. As far as I can see, fa koina is never used in the SE, as it is in the Post. Analylics, to pick out the special class of principles of demonstration, such as the law of excluded middle, that underlie demonstrations in the various sciences; cf. A Po 88a36-bl . (37) Cf. SE I70a34-36 with bS-1 1 ; see also 1 7 lb6-7, I72a29-3! , Rhel. !358a!0-17, 1354a!-3.

280

THE EPISTE,MOLOGICAL BASIS OF ARISTOTELIAN DIALECTIC

(!Ol a30-33). Hence to make use of the koina is to make use of generally held opinions. (Cl. Bolton, 46-7; but the generally held opinions referred to in Topics I 2 are clearly not "things commonly known to be true".) There are other passages, however, in the SE and the Rhetoric, where the term koina designates common "lopoi": i.e. inference patterns or premises common to several sciences or disciplines.38 Consider the following passage from SE 9: Clearly, then, it is not of all refutations, but only of those that depend upon dialectic that we need to grasp the lopoi; for these are common (koinoi) to every art and faculty. . ..Accordingly it is clear that it belongs to the dialectician to be able to grasp on how many considerations depends the formation, through the koina, of real or apparent refutations.39 The fact that dialectical (or peirastic) arguments are said to make use of common lopoi gives us no reason to suppose that the premises of such arguments must be true. An example of a common lopos is "the more and the less" .•• Under the rubric of this rule or inference pattern, Aristotle suggests a number of strategies for arguing that one thing is more or less F than another.41 These strategies can be used in widely different domains, e.g. in discussions of physical as well as ethical

questions; this is why the lopos of the more and the less is "common". These strategies may be used not only in dialectic but in the sciences as well; when they are used in dialectical arguments, the premises may be true but they need not be: the essential requirement is that they be plausible.42 The common lopoi are "common" in yet another sense in that they are employed not only by trained dialecticians, rhetoricians and scientists, but also by ordinary people with no training in these fields. This is why Aristotle says that laymen "know these things", i.e. the common lopoi, no less than trained experts .43 He is not saying that the arguments of laymen are based on premises known to be true, but that laymen-no less than experts-know about and make use of certain common strategies of argumentation. Reliance on koina is a feature shared by peirastic, dialectical and rhetorical arguments, as well as by arguments of untrained laymen. We have seen that there is no basis for viewing the premises of peirastic arguments as truths that are matters of common knowledge and belief. The fact that peirastic arguments make use of certain koina, i.e. certain lopoi, does not imply that their premises must be true, or even that they must be endoxa. They may be endoxa, but they need not be. What is essential is that they be beliefs held by the answerer, for only if this is the case can there be a genuine test of the answerer's claim to know.44

·

i' 1

281

D. DEVEREUX

(38) An example of a lopos in discussions of comparative value would be "the most conspicuous good is more desirable than the less conspicuous, and the more difficult to attain than the easier" ( 1 1 7b2-29; cf. altos in 1 1 7b28 with allos lopos in 1 1 7b10). A lopos could be viewed as a rule that guides the formulation of an argument without actually appearing as a premise in it; but there is no reason why a lopos like the one cited could not be used explicitly as a premise in an argument. Sometimes Aristotle speaks of topoi as premises of dialectical or rhetorical arguments; Rhel 1358a2-32. An important part of Aristotle's conception of a lopos is that it is a matrix from which one can generate arguments appropriate for different circumstances (119al-14 ; cf. 105b3-37, 163b3-164a 1 1 , 164b1-19}; a mastery of the lopoi will enable one to discover the premises one needs in a given discussion, and this is the mark of a skilled dialectician (164b3).. Aristotle's predecessors had their students memorize question and answer debates; in so doing, they were imparting the products of the art but not the art itself (SE 183b3-184a8). Perhaps more than anything else, it is the notion of a lopos as a sort of matrix of a_rguments that distinguishes Aristotle's teaching from that of his predecessors. (It is interesting to note that one meaning of lopos is the female genitalia: see L. S. J., lopos, I 3. But Aristotle may have been influenced more by the way the term was used in contemporary discussions of mnemonic techniques; see Top. 163b23-33.) (39) 170a34-36 and bS-10; cf. 1 7 l b6-7, 1 72a29-3 1 , Rhel. 1358al0'.\7. Bolton (n. 30) recognizes that koina is used in this sense in the Rhetoric, but thinks it is not used in this sense in SE 1 1 . But in the passage we have just quoted from SE 9 it is clear that the term does not designate generally held opinions; the use of koina here is clearly parallel to the use of the term in Rhel. 1358a10-17. Furthermore, the parallel between 170a34-36 and 172a29-30 indicates that the term is used in the same sense in the relevant passages in SE 1 1 . Koina seems to be used only once in the Topics {164a8), and there it seems to be a synonym or "universal" (kalholou). (40) See Rhel. 1358a10-17 and Topics I I I , passim. (41) These particular strategies are also called lopoi ; cf. 1 1 7b10ff.

(42) See, e.g., I l l 6, esp. l l9a38-bl , 1 19bl5-16. (43) SE 172a32-34; cf. Rhel. 1354al-6. (44) See above, pp. 273-4. SE 183a37-bl ("Our aim was to discover a faculty of reasoning about any question put before us on the basis of the most plausible views available. For this is the function of dialectic in the strict sense (kath' hauten) and of peirastic.") seems to imply that the premises of peirastic as well as dialectical arguments must be endoxa (see Bolton, 200). But we should note, in the first place, that the activity referred to at 183a37-38 is not the whole function (ergon) of dialectic; he refers only to the activity (or ability) of the dialectical questioner, and then in 183b3-6 he mentions the others task of "upholding a thesis", i.e. playing the part of answerer in a dialectical discussion. The fact that Aristotle says that the first activity is the function of "dialectic kafh' hauten" seems to indicate that it is the ability to question that is most characteristic of the dialectician. Is Aristotle saying (as Bolton apparently thinks) that it is the function of peirastic and of dialectic in the strict sense to argue as questioner from what it most plausible, while it is the function of dialectic in a looser sense-and not the function of peirastic-to uphold a position as answerer? The problem with this is that it is difficult to see how peirastic understood in this way fits with what is said about peirastic earlier in the SE; here "peirastic" seems to designate the general ability of a dialectician to conduct a test (peiran labein, 183b2-3) through questioning, whereas earlier the term was used for the special ability to unmask false knowledge claims. It is worth noting that in ms. D the article tes before peirasfikils (in 183a39- b l ) is missing, so that peiraslikils functions as an adjective modifying dialeklikils. Now there .are other instances where peirastikil is used as an adjective modifying dialectic; cf. Metaph. I004b2-26, SE 171 b9. What is it for dialectic to be "peirastic"? 183b2-3 gives us a clue:

282

D. DEVEREUX

The aim of peirastic arguments, as we have seen, is to unmask false claims to knowledge. Such arguments do not rely exclusively on endoxa, and are therefore not a subspecies of dialectical arguments. Peirastic arguments are related to dialectic in that the ability to conduct a peirastic examination is part of the dialectician's art. We have noted that it is also by virtue of dialectical skill that one is able to argue sophistically: the dialectician could, if he chose to, perform very effectively as a sophist; and the sophist, if he chose to, could hold his own in a dialectical discussion. Peirastic and ·sophistic are two different ways of exercising dialectical skill�one perfectly acceptable, the other not acceptable.45 In his scattered remarks concerning peirastic examination, Aristotle seems to be thinking of Socrates' method of testing knowledge claims and trying to locate this method in relation to his own conceptions of dialectic and sophistic. We have noticed several indications that he is not thinking of peirastic as a form of dialectic that might be useful in a scientific or philosophical investigation; the premises of peirastic arguments need not be true, nor do they need to be plausible. The upshot is that neither in the Topics nor in the SE does Aristotle envision a kind or method of dialectic that would have some essential role in the sciences or in philosophy.

I

I' I

!

insofar as dialectic is able to "conduct a test" (peiran labein), it is peirastike. In other words, describing dialectic as peirastike is a way of focusing on the function of conducting a "test" (peira) through questioning. If we adopt the reading of D, we no longer have a problem reconciling 183a37-bl with earlier passages concerning peirastic. This passage does not say anything about peirastic understood as the ability to unmask false claims to knowledge. Aristotle is simply referring to the two roles of the dialectician, questiQning and answering or "conducting and submitting to a test", and he assigns priority to the first. Another reason against taking 183b2-3 as referring to peirastic examination is the general context. 183a37-b13 is a resume of the subjects discussed in the Topics; it is only at b l 3-15 {cf. 183a27-36) that he mentions the contents of the SE. {Cf: Brunschwig, Topiques, XIX, n. 5). Since the Topics contains no discussion of peirastiC examination and no use of the expression peiran labein to refer to such an examination, it would be out of place to allude to this type of activity in a summary of the contents of the Topics. If it is possible to understand peiran labein in 183b2-3 as referring to an activity discussed in the Topics, and also to understand peiraslikes in b i in a similar way, clearly this would be preferable in view of the general context. And if the passage is understood in this way, there is no implication that the premises of peirastic arguments {as defined in SE 2) must be endoxa. Of course, if peirastic premises express beliefs held by the answerer, they must be endoxa in the minimal sense that they are plausible to him; cf. Top. 159a38-b4, 159b227. But it was not endoxa of this sort that were originally said to be the premises of dialectical arguments (Top. 100a29-b23, I 10). {45) Perhaps this explains why peirastic is said to be a "part" of dialectic whereas sophistic is not; when a dialectician conducts a peirastic examination he is still acting qua dialectician, but when he meets the sophist on his own ground he is not.

f

THE EPISTEMOLOGICAL BASIS OF ARISTOTELIAN DIALECTIC

283

IV The position we have argued for thus far seems to fly in the face of Aristotle's explicit statements about the usefulness of dialectic in relation to philosophy in Topics I 2. Particularly troublesome for our interpretation is his claim that the first principles of the sciences must be treated on the basis of endoxa, and that this is the · proper task of dialectic. Does this not imply that dialectic has an indispensable role in the sciences in regard to the discovery and establishment of first principles? In spite of the strong attraction of such a view, I do not think we should take Aristotle's statement about dialectic and first principles as an indication that one of the aims of his treatise is to formulate a method to be used for the discovery or "validation" (Bolton) of the first principles of the sciences. In the first place, as we have seen, there is no trace of such a method in the Topics or SE. Secondly, and more importantly, in a number of passages we find a sharp distinction between the activity of the dialectician qua dialectician and the activity of the philosopher. When we are engaged in a dialectical inquiry or discussion, we are not doing philosophy, and vice versa.46 Hence the philosophical utility of dialectic must be understood in some way other than through its providing a method to be used in a philosophical inquiry. It might be useful to consider the relationship between dialectic and sophistic. The activities of the sophist and dialectician qua dialectician are very different: someone engaged in a sophistical debate is not carrying on a dialectical discussion or inquiry, and vice versa. Nevertheless, the sophist uses dialectic in the sense that the skill he employs is the same as that of the dialectician; there is not a separate skill of sophistic corresponding to sophistical arguments. 47 So training in dialectic is useful to the aspiring sophist, not by providing a special (46) See. e.g., 155b7-16, 105b30-31 ; cf. 162b31-33, SE I75a31-33. According to 155b10, a distinctive feature of dialectical discussion is that it is conducted "in relation to another" (pros heleron). 'However, at 163a36-b4 Aristotle says that we can practice dialectical argumentation by ourselves, and this is sometimes cited as evidence that it is not an essential feature of dialectic that it involve a discussion between two people (a philosopher working by himself can be arguing dialectically). But the sort of practice Aristotle is thinking of at 163a36-b4 would still be done "in relation to another": we would simply imagine ourselves in first the role of questioner and then the role of answerer; the distinction between dialectical and philosophical argument noted in 155b7-16 \vould apply no less to such solitary dialectical discussions. {47) See above, n. 29.

!

r1 I

I '

284

D. DEVEREUX

method, but by giving a skill in formulating and analyzing arguments which is transferrable to sophistical debates. Dialectical skill is thus useful in non-dialectical as well as dialectical contexts." In the initial stages of a philosophical inquiry (a non­ dialectical context), Aristotle believes it is useful to "problematize" (diapore6)-to formulate arguments on both sides of various controver­ sial issues. These arguments will not normally be dialectical, but the ability to formulate them is a matter of dialectical skill." This is the first of the two ways in which dialectic is said to be useful to philosophy m Topics I 2. The second way in which dialectic may be of use to philosophy is described in the following passage. It has a further use in relation to the principles used in the several sciences. For it is impossible to discuss them at all from the principles proper to the particular scien?e in ha!1� · seeing that the principles are primitive in relation to everything else: 1t 1s through the endoxa about them that these must be treated, and this task belongs properly, or most appropriately, to dialectic; for insofar as it tests, it provides a way (hodos) to the principles of all inquiries. This passage clearly seems to imply that it belongs to dialectic to discuss the first principles of all branches of knowledge on the basis of endoxa. But let us note, first of all, that the passage is doubly anomalous. As Brunschwig has pointed out, it seems to fall outside the classification of ways in which dialectic is useful which we find at the beginning of chapter 2.50 More importantly, there is no reference to this very important task of dialectic anywhere else in the Topics-or indeed anywhere else in Aristotle's works. There are well known passages about how we come to know first principles; there are even references to "the way" (hodos) to the first principles of the sciences.51 But in none of these contexts is there any mention of dialectic or of the need to discuss principles on the basis of endoxa. The claim that it is necessary (anangke) to discuss first principles on

I,

'i ' '

,I

(48) One example of such a non-dialectical context is the day to day encounters (enleuxeis) that are mentioned at 10l a30-34; cf. Rhet. 1355a21-29. Insofar as rhetoric depends on dialectical skill (cf. 1356a25-27, 1359b9-12), speaking in the public assembly would be another non-dialectical context in which dialectical skill might be exercised. (49) In Melaph. B, for example, there is no indication that the various arguments formulated pro and con rest exclusively on endoxa. (Note that at 995b20-25 Aristotle seems to imply that what the dialecticians do is different from what he is doing.). (50) See Topiques, 1 1 6-7. It might be argued, against Brunschwig, that what Aristotle means by las kala philosophian epistemas in 1 0 l a34 is not a subclass of hapasOn ton melhodOn in 10Ib3. But does Aristotle believe that raising difficulties on both sides of an issue is useful in all areas - e.g., in mathematics? At A Pr 46a3-4 Aristotle distinguishes between philosophical and other types of inquiries. (51) See, e.g., A Pr. 43a20-22, 46a3-27, A Po. 84bl9-31, I I 19.

]i

1 !1

i

l

THE EPISTEMOLOGICAL BASIS OF ARISTOTELIAN DIALECTIC

285

the basis of endoxa also seems to conflict with Aristotle's view in the Post. Analylics. According to I I 8-10, we come to know some definitions, e.g., the definition of lunar eclipses, through scientific inquiry. We discover that an eclipse of the moon is caused by the earth's blocking the sun's light, and this is then incorporated in our account of what an eclipse is. A definition of this sort would presumably count as a principle;" and it is clear that the way we arrive at it is not through dialectical discussion. Aristotle's understanding of how first principles in the sciences are arrived at in the Analylics seems incompatible with the sweeping claim about dialectic in our passage in Top. I 2. Arguments based on endoxa would seem more appropriate in areas where empirical investigation is of little or no help in the discovery of first principles, e.g., in first philosophy or ethics. But the claim in I 2 applies to all branches of knowledge, and in any case it is difficult to see how dialectical discussion, as described in Topics V I I I , could provide a way to the discovery of true first principles. In view of the fact that Aristotle's sweeping claim about dialectic seems to be anomalous in its context, we might consider the possibility that it is a vestige of an earlier, more "Platonic", version of the Topics.53 In Book V I I we find the suggestion that it is possible, though very difficult, to "establish" (kalaskeuazein) definitions through dialec­ tical arguments -" Throughout the central books of the Topics ( I I- I I I) Aristotle speaks of "establishing" or " destroying" such claims as that X is the genus of Y, or that X is "proper" (idion) to Y . He may mean "establishing" or "destroying" according to opinion only (kala doxan), so that even though one has "established" that X is the genus of Y, in truth it might not be. However, as Moraux suggests,55 one gets the impression in these passages that Aristotle means something stronger. Nowhere in I I-VI I do we find the contrast between arguing kala doxan and kal' aletheian, and when Aristotle speaks of establishing some claim he does not give any explicit indication that "establish" (52) See, e.g., 158a31-b4, 158b24-159al . (53) llow would the "vestige" find its way into the later version? We might imagine that a later editor had two texts of this part of the treatise: one an earlier, the other a revised version; wherever the earlier version had something that seemed important to the editor, but was lacking in the later version, he would insert it into the latter. An editor of Aristotle's works would naturally be reluctant to discard anything written by the master. (54) See V I I 3 and 5. Aristotle does not claim that one can "demonstrate" a definition, but only that one can "syllogize" or deduce a definition; thus there seems to be no explicit contradiction between these chapters and A Po I I 3-7. (55) Op. cil., 307-8. I would not agree with Moraux's suggestion that dialectical arguments in II-VII are not understood as arguments resting on endoxa; it is rather that Aristotle believes at this point that arguments resting on endoxa may yield conclusions that are true and well founded. Cf. Rhel. 1355a14-15.

l :' ,

i

286

D. DEVEREUX

must be understood in a qualified sense. If he believed, at the time of writing Book VII, that one could establish through dialectic that a certain formula expresses the true definition of something, and that definitions are principles of the various sciences,56 then of course it would not be surprising to find him making the sweeping claim about dialectic that we find in I 2.57 Whether or not one finds this hypothesis plausible, the fact remains that IO!a36-b4 has no echo elsewhere in Aristotle's works, and actually seems to conflict with a number of passages that discuss the "way" to first principles. It would be imprudent to suppose, on the basis of this single passage, that Aristotle's mature conception of scientific and philoso p hical inquiry assigns an important role to dialectic.

"

i

!\ I

BIOLOGIE, LOGIOUE El' MlttAPHYS!QUE chez A1usroTE

8eminaire CNRS-N.S.F., 1987

Editions du CNRS, Paris, 1990.

SCIENTIFIC EXPLANATION AND EMPIRICAL DATA IN ARISTOTLE'S METEOROLOGY1

* **

I will conclude with a few caveats. My brief remarks concerning the question of the philosophic utility of dialectic obviously do not settle things one way or the other; at best, they will raise doubts about whether Aristotle's claim at IO! a36-b4 reflects his mature view of the relationship between dialectic and philosophy. My chief aim in the essay was to show that there is no evidence in the Topics and SE for a special form of dialectic to be used in philosophical or scientific investigations; it is possible that the remarks about the usefulness of dialectic in Topics I 2 have in view a special dialectical method which is described in another work.58 I must content myself with having shown that there is no sign of such a method in Aristotle's treatise on dialectic.

' •,

•i !

(56) The Topics does not mention any other kinds if principles besides defin�tions. (57) I t is in its capacity as exelastike that it provides a way to first principles, so

perhaps we should understand the procedure of establishing a definition as including showing that other proposed definitions are unsatisfactory. (58) For example, it is often claimed that the method described in N�c. Ethics V I I 1 is the method of dialectic. But Aristotle does not use the term "dialectic" in discussing that method, and as far as I Can see there is no connection between it and what is described as dialectic in the Topics and SE.

I

l

i1 I

CYNTHIA A. FREELAN D

INTRODUCTION In this paper I intend to examine certain significant aspects of . Aristotle's scientific practice as recorded in his Meteorology.' I will survey a number of examples in order to study the complex relations between scientific theory and empirical data in this treatise. Aristotle advances definitions of a host of meteorological phenomena; these defimtwns, very broadly considered, conform to the model recommen­ ded in the Posterior A nalylics.3 That is, individual meteorological . subjects such as comets or thunder are defined in terms of a causal analysis resting upon a unified set of underlying principles. The first (1) An earlier version of this paper was presented to Robert Bolton's seminar on Aristotle.'s. philos�phy of scie� ce at Rutgers University in March, 1987. I am grateful to the part1c1pants 1n the seminar, and in particular to Tim Maudlin, for questions and comments. In general I am much indebted to Bolton for his comments on these topics in �umerous con�ersations over the past several years. I also hope that the paper has been improved by incorporating responses to questions raised when the earlier version was delivered in La Rochelle. (2) In fact I will only discuss Books I - I I I of the Meteorology. This is due less to c�ncerns abo.ut the au�henticity of Book IV than to its fairly obvious lack of continuity with the earlier books 1n terms of approach and methods. The absence in Book IV of the fundamental arc?ai, the exhalations, consistently used throughout I-III, makes Book IV �eem part of a different project (although it covers subject-matter we would expect to be included on the basis of the treatise's overall program). See also my discussion in Fceeland (1986). (3) See Post. An. I I , 10 and for an interpretation of this model of "real definitions" see Bolton (1976). '

i:i ,

,

288

C. A. FREELAND

principles of this science are set forth in Book I, 3; they are two fundamental exhalations, one hot and dry, one cool and moist. I will not here pursue important questions concerning these first-principles in particular whether this treatise employs the deductive explanatory structure recommended in the Posterior Analytics.4 Instead, my focus will be on certain steps recorded in this treatise which are not mentioned in Aristotle's model of pure science - steps which precede and follow upon his core explanations (definitions) of meteorological phenomena. In Part I, I will explore the role of the endoxa, the · reputable opinions or earlier scientific beliefs which Aristotle surveys before providing his own accounts of various phenomena in this science. The proper assessment of Aristotle's use of endoxa in science is a topic of much current interest and debate.5 I believe that the surveys of endoxa play a dual role in this work, enabling Aristotle to refine why-questions and to gather relevant empirical data or phainomena. The extent of their role cannot be fully grasped without a correlative understanding of Aristotle's use of "signs" (semeia) and "proofs" (iekmeria), which are often mentioned in his arguments for the correctness of his own accounts. I n Part I I , I will show that Aristotle uses signs and proofs in an inference pattern now termed "abduction" , to argue from explanato­ ry success to the truth of his proposed analyses. In my concluding Part I I I , I will address another topic that has been of recent interest, whether Aristotle means his account of scientific knowledge in the Posterior A nalytics to serve as a model for the cognitive state of a scientist prepared to provide explanations, the best structure for organizing and imparting scientific knowledge, or a recreation of the processes of real scientific inquiry .6 I argue that in the key respects in which Aristotle's practice in the Meteorology differs from the proposed model of the Posterior A nalytics, he adds more than mere rhetorical supplements to the ideal model of the latter text. Instead, he reveals a concern for what contemporary philosophers of science call the "pragmatics of explanation . "

f

SCIEN'l'IFIC EXPLANATION AND EMPIRICAL DATA

289

Part I. The Role of Endoxa within the Meteorology

A. OVERVIEW As defined in the Topics, the endoxa include "what seems so to all or the majority or the wise" (!00b21-4). Aristotle provides no special rationale in the Meteorology for his frequent surveys of predecessors' views, unless we are to count his remark at 339b28-30: We maintain that the same opinions recur among men in rotation, not once or twice or occasionally, but infinitely often.1 In general in his surveys Aristotle studies only the opinions of the wise - Anaxagoras, Heraclitus, Empedocles, the Pythagoreans, Demo­ critus, Hesiod, Hippocrates of Chios, Anaximenes, Cleidemus, and Plato. In no case does he treat any of these men as having offered a unified and comprehensive theory of the meteorological realm, a full rival to his own. Instead, in the course of laying out his own remarkably unified theory, he proceeds via an apparently random series of attacks on various aspects of his predecessors' accounts. In just one instance does Aristotle seem impressed by an earlier view - an account of rivers and winds is termed " ingenious" (kompseuma) at I, 13, 349a283 1 . More typically, his predecessors are quickly, even derisively, dismissed. Thus Aristotle complains that previous accounts of the winds might have been proposed by (in Lee's translation) "the man on the street." Democritus' explanation of why the sea is salty is compared to one of Aesop's fables ( I I , 3), and Empedocles' account of the same phenomenon is relegated, as metaphorical, to the realm of poetry. Again on the topic of winds Aristotle criticizes an earlier view, saying, "The unscientific views of ordinary people are preferable to scientific theories of this sort" ( I , 13, 349a27-8). He often claims not only that previous theories are false but that they are "silly" or "absurd." B. DIALECTICAL SURVEYS OF ENDOXA

I :

(4) I d o this i n a companion piece, Freeland (1986). O f course, separating these questions from the subjects of this article will result in some regrettable artificiality of analysis. (5) See Barnes (1980), Bolton (1987), Brunschwig (1987), Cooper (1988), Devereux (1987), Kakkuri-Knuuttila ( 1987) and (1988), Nussbaum (1982) and (1986), and Owen (1961). (6) On these debates, see in particular Burnyeat (1981) and Barnes (1975).

'

: .

iI

I

1· I

!

I

L

Aristotle also dismisses proposals by saying that they lead to " impossibilities" (adunata). The implications of this charge fluctuate. Occasionally he says that the proposed explanation can be demonstrated to be inadequate by the use of reason alone. Thus, for (7) The Greek text cited throughout is that of the Loeb edition (Cambridge, Massachusetts: llarvard University Press, 1982), and the translations are those of the Loeb translator, H. D. P. Lee, with occasional modifications.

! I i!

290

'

! !

i

C. A. FREELAND

example, he criticizes Democritus' view that comets are (appearances resulting from) conjunctions of planets by arguing that stars (here including planets) appear as individual points; but the appearance of the conjunction of two individual points would itself be an individual point, not a comet or something with any greater magnitude. Aristotle says then that this theory fails on the basis of logic or reasoning (ek lou logou). 8 There are other similar instances in which Aristotle rejects a proposal by a relatively simple process of reasoning a bout its consequences. Thus for example in I, 13, he considers what must be the case if we accept the theory that rivers originate from the collection of rain within a reservoir deep inside the earth: . .. It is evident (phaneron) that if anyone tries to compute the volume of water constantly flowing each day and then to visualize � re�ervoir for it, he will see that to contain the whole yearly flow of water 1t will have to be as large as the earth in size or at any rate not much smaller. (349bl6-19) Again, he argues in I I , I against the endoxon that the sea has sources, as follows: ... [I have given] a complete enumeration of the various species of water: and from this classification one can see that it is impossible (adunalon) for the sea to have sources. For water that has a source is either running or artificial: but the sea has neither of these characteristics... And we know of no volume of water of comparable size that has sources and yet stands of its own accord. (353b30-35 passim) Later, in I I , 9, Aristotle challenges Empedocles' and Anaxagoras'. view that thunder and lightning result when fire is trapped in clouds by noting, . ..There must be a separate and distinct cause of the occurrence of each phenomenon, whether thunder or lightning or anything else . Bu.t the cause proposed is far from fulfilling this requirement. It is rather as if one supposed that water and snow and hail emerged ready-made, and did not have to be formed because the atmosphere has a stock ready to hand for each occasion. (369b28-34)

{8} Among points disputed in recent discussions is whether the Topics characterizes dialectic, a special form of logical examination of beliefs through question and response, as itself "the method of endo::ca" or as a distinct enterprise with its own rules. Part of what is at issue involves the Topics' fairly rigid distinction between roles of questioner and answerer. Another issue coneerns whether dialectical inquiry can "test" the endo::ca against empirical data. S.ee Bolton (1987}, Brunschwig (1987), Devereux (1987), and Kakkuri-Knuuttila (1987}. In describing two major types of treatments of the endo::ca in the Meleorology, one generally "logical" and one "empirical," I am not committing myself to a particular view on these disputed issues. However, I do think that the particular way in which Aristotle tests the endo::ca against empirical phainomena in this treatise is not envisaged by the Topics; it goes beyond even Bolton's characterization of how dialectic (in particular "peirastic"} can help to supply empirically adequate explanations. See my section IC, below.

I, ,

r

SCIENTIFIC EXPLANATION AND EMPIRICAL DATA

291

In another especially interesting instance of what might be classified as an ordinary " by-the-book" dialectical refutation of a proposed explanation,9 Aristotle advances a three-pronged attack in I I , 2 o n the Heraclitean view that the sun i s fed b y moisture. He argues first that if this is true, then the sun would be constantly changing, like a flame: "Clearly there would not only be, as Heraclitus says, a new sun every day, but a new sun every second" (355a l3-15). Second, there is no more reason to believe that the sun is fed by moisture drawn off the earth than to think that a fire is fed by the water heated over it. And third, the Heraclitean fails to explain the reciprocity between moisture rising up and what returns to the earth; but as Aristotle says, .. . We can see clearly (phaner6s hor6men) that the water drawn up always falls again. Even if the correspondence is not exact in any one year or any one place, yet in a certain fixed period what was taken is retur­ ned. (355a26-28) To summarize, in these sorts of discussions of endoxa Aristotle reasons against an opponent on the basis of certain unquestioned assumptions (such as that stars appear as points), simple classifications (such as the account of the species of water in I I , !), or common-sense inferences (such as that the earth could not contain a hidden reservoir of water that is nearly its own size). These discussions are not especially remarkable from the point of view of standard descriptions of dialectic. But they are not typical of Aristotle's treatment of endoxa in this treatise. For the most part, Aristotle tends to refµt.e the endoxa in this science by placing them up against certain phainomena or observed facts - what we might call the empirical data. Indeed, this is most frequently the method by which he presents the data which are of special concern and which an adequate scientific explanation must take into account. C. ENDOXA AND EMPIRICAL PHAINOMENA At times Aristotle begins by listing a few key facts about some subject, as he does concerning hail, for instance, in I, 12: In considering the process by which hail is produced, we must take into account both facts whose interpretation is straightforward and those which appear to be inexplicable. (347b34-37) More often, though, Aristotle launches straightaway into his examination of the endoxa, or at any rate of the strongest existing explanatory candidates, and he proceeds to show their inadequacies. His critique may mention some prediction the proposed account makes which turns out to be false, some fact it has overlooked, (9) For a general outline of Aristotle's account of dialectic in the Topics, see Kakkuri-Knuuttila' (1987).

, i:

·1'.• I SCIENTIFIC EXPLANATION AND EMPIRICAL DATA

or simply some claim it makes which does not accord with the facts as "we have observed" them. I should emphasize that Aristotle does not make use of surveys of the endoxa in connection with each and every topic in the Meteorology, but rather with those he finds m@t puzzling. These include, for example, comets (I, 6 and 7), hail (I, 12), the origin of winds and rivers (I, 13), the saltiness of the sea ( I I , 1 and 3), earthquakes ( I I , 7 and 8), and thunder and lightning ( I I , 9). I will next examine two of the most interesting of these studies, those c_oncerning comets and hail, in more detail.

Aristotle concludes by remarking, "Though more could be said, this is enough to demonstrate the falsity of current theories of the causes of comets" (344a2-4). Notice exactly what has occurred in Aristotle's discussion of the endoxa concerning comets. First, a number of empirical facts (phaino­ mena) have emerged which were not included in the initial characteriza­ tion (nominal definition) of comets. We have learned that, like stars, comets are bright bodies which emerge and follow certain paths through the sky; but they differ in their actual appearance - in having tails. Comets do not occupy the general region of the sky occupied by planets; nor do they occur in any fixed constellation, season, or region of the sky. They seem to appear and disappear at random, and sometimes more than one comet appears at a time. Second, Aristotle has done a great deal to focus his inquiry by raising very specific questions: Why don't comets follow the zodiac? Why do they sometimes appear with stars, sometimes not? Why are their appearances more random and infrequent than those of stars and planets? Third, having criticized earlier views for failing to answer these questions about comets, Aristotle is now under a prima facie obligation to propose an account which will explain the phainomena just noted. That is, in addition to explaining the most obvious feature of comets, their tails, Aristotle must now account for their positions and movements 1 generation and dissolution, relation to stars and planets, and infrequency of appearance. I will briefly sketch Aristotle's own account of comets in Section F below; but I am not really so much interested in assessing his actual explanation as in underscoring Aristotle's own efforts to argue for his theory's explanatory adequacy and superiority to its rivals. These arguments will become clearer when I review his use of signs and proofs in Part I I , below.

D. COMETS Aristotle devotes all of Meteorology I, 6 to a discussion of previous accounts of comets, before moving ahead to give his own explanation in I , 7. Here is a sketch of each of the views he considers: (i) Democritus and Anaxagoras say that comets are conjunctions of planets; they appear to touch each other because of their nearness. (ii) Some Pythagoreans say that a comet is one of the planets which (like Mercury) appears seldom, and never far above the horizon. (iii) Hippocrates of Chios says that a comet' s tail is an appearance caused by reflective moisture drawn along by the comet's motion. It only acquires a visible tail when seen in the north. In essence, here are Aristotle's responses to each theory:

i

I

I

293

C. A. FREELAND

292

Against (i), (effective against (ii) and (iii) as well) : (a) Some of the fixed stars have tails. "And for this, we need not rely only on the evidence of the Egyptians who say they have observed it; we have observed it also ourselves. For one of the stars in the thigh of the Dog had a tail, though a dim one: if you looked hard at it the light used to become dim, but to a less intent glance it was brighter." Against a variation of (i) which holds that a comet can result from the conjunction of a star with a planet: (b) In this case, stars should always appear at the dissolution of a comet, but this is not so. And furthermore, "We ourselves have observed the planet Jupiter in conjunction with one of the stars in the Twiµs and hiding it completely, but no comet resulted." (c) "On purely logiCal grounds" we can see that the conjunction of two stars (or of a star and a planet) which each appear as points will itself appear as a point, not as a comet [this case was reviewed just above]. Against (ii) : (a) The planets all fall into retrogradation within the zodiac circle, but many comets have been seen outside the circle. (b) More than one comet has been seen at a time. Against (iii) : (a) Sometimes this planet should be seen without its tail (which has the reflective appearance), but it is not. (b) It's not true that comets appear only in the north, and only during the summer solstice; Aristotle cites instances of comets seen in the south and west, and in the winter solstice.

E. HAIL Aristotle proceeds differently in his discussion of hail; he begins by listing various phainomena, some of which he calls straightforward, others hard to explain (paraloga). He assumes a common understan­ ding of the nature of hail (or of its nominal definition): "For hail is ice, and this is obvious to all" (348a33); also obvious, of course, is that hail is ice which falls from clouds. But beyond these obvious facts, Aristotle cites the following additional sumbainonla: (i) Hail is ice, yet it's commonest in spring and fall, less so in summer, and least in winter, the coldest season; similarly, it falls in milder districts. (ii) Hail seems to be water frozen up high in the clouds; yet this is odd, for the water has to be up there to get frozen, yet once it's frozen it can't stay suspended for very long. And hailstones can ' t grow in size while they ' re up in the clouds, since frozen drops won't coalesce as liquid ones will. (Cf. 347b36-48al4)

I

,,

'I i

I,

i

II

!

I

I

I

II

II

11

I I

i!

"

1 Ii "

I

,[ i

I

1' 1

294

C. A. FREELAND

Next Aristotle considers a previous theory - only one is deemed worthy of note, that of Anaxagoras. Anaxagoras held that hail is formed when a cloud gets forced into a cooler region up above it, where the sun's warmth has not been reflected up from the earth. "And so hailstorms occur more often in summer and in warm districts, because the heat forces the clouds up further from the earth" (348al821). Notice that Anaxagoras' theory "explains" not only why hail falls and how it is formed, but why it falls when it does. Neve_rtheless, Aristotle raises problems about its explanatory adequacy (348a2137). He notes that according to Anaxagoras it should hail more in high places - but it doesn't. Second, sometimes the biggest hail falls from clouds closest to the earth, but on Anaxagoras' view we would expect just the opposite. Finally, Aristotle observes that sometimes hail is very large but irregularly shaped, indicating it has fallen close to the earth (if they had formed higher up, the hailstones would have been rounded off by their fall); but once again, this is not what Anaxagoras' view would lead us to expect. Here again, as in the case of comets, Aristotle introduces the phainomena or empirical data about hail via his critique of an endoxon. In addition to knowing the facts initially cited in I, 12, we now know that hail falls a lot in low lands and from clouds near the earth - indeed, this is the source of the largest hailstones. Sometimes hailstones can be very large but not rounded. Once again, Aristotle has raised quite specific questions about his subject: Why does hail occur more in the fall and spring than winter, when cold temperatures are more prevalent? Why do larger hailstones fall from clouds nearer the earth, when it would seem that if they fell further they could get larger? What accounts for large irregularly shaped hailstones? Once more Aristotle has committed himself to explaining these complex constellations of facts. He must account for the locale and seasons of hailstorms, as well as certain facts about the sizes and shapes of hailstones falling from particular types of clouds. F. ENDOXA AND ARISTOTLE'S EXPLANATIONS Aristotle may at the very least use the endoxa to supply a nominal definition at the start of an inquiry; certainly he considers it necessary to account for why there are "comets", i.e. starry objects with fails, or why there is "hail", i.e. ice which falls from the sky. But it is evident that Aristotle feels no obligation here to propose a theory that "explains" all the endoxa on a given subject, even those he considers mistaken.10 He

(10) This runs counter to what many see as his attitude in a methdologically important section of the Nico1nachean Ethics. See, for example. Cooper (1988), p. 553:

l

SCIENTIFIC EXPLANATION AND EMPIRICAL DATA

295

would regard it as misguided to concoct some theory explaining Anaxagoras' mistakes about hail or Democritus' mistakes about comets. Aristotle at least credits his predecessors with having been engaged in the same sort of enterprise he is; so glancing back at their efforts can help his own inquiry both by raising new questions and providing new data. Surveys of endoxa furthermore set up expecta­ tions for explanatory adequacy. Aristotle aims to generate an account that will be more than merely consistent with the empirical facts, but one that will explain the phainomena concerning comets, hail, and so on, many of which have emerged as the result of his critical examination of the endoxa. He does not go so far as to maintain explicitly that his own scientific accounts will permit a strict derivation of all the plainomena he has recorded concerning comets, hail, etc. N evertheless 1 his practice indicates that he is confident his own proposals afford greater explanatory power than their rivals, and he takes pains to point this out.11 A rough overview will enable us to see whether and to what extent Aristotle's own explanations of comets and hail constitute improve­ ments over those of his predecessors. The key components of all the scientific explanations offered in Meteorology I-III are the two archai which are basic in this science. These serve primarly as material causes of various phenomena in this realm, though they also seem at times to possess efficient-causal force. In his account of comets Aristotle begins by alluding to what he has hypothesized earlier (hupokeilai) concerning the hot dry exhalation. He then proceeds to explain the nature of comets by alluding to this exhalation and also to the movements of

"Why ... does Aristotle make philosophy answerable even in this limited way to the endoxa? In his extant works he never addresses this question squarely, but the obvious answer... is that there is antecedent good reason to expect that some or all of what is reputably believed is true.. Other things being equal, intelligence when applied to a problem gets things right ... " Cooper is here addressing claims made by Nussbaum (1986). See also Bolton (1987), especially Section IV, p. 35, for discussion of Aristotle's views on the relation between endoxa and truth. Brunschwig (1987) helpfully suggests that there may be some divergence in Aristotle's opinions on this issue as expressed in his ethical writings and his natural scientific works. (I I) Owen (1961) argues that a successful theory must leave the phainomena standing; to this it may be replied that Aristotle sometimes expects more of a theory that it not merely be consistent with the phainornena but actually explain them. See on this Bolton (1987). There are complex issues here about whether the phainomena include endoxa or are separate, referring rather to "empirical data" more narrowly understood. I cannot go into these general issues here. See, again, Owen (196I), Bolton (I987), also Nussbaum (I982). I think it is evident in the Meteorology that Aristotle holds that an adequate theory must explain, and not merely be consistent with, the empirical date. But for one puzzling remark running counter to this interpretation see my Appendix I r below.

I' ,'

f

I

'I

'

296

SCIENTIFIC EXPLANATION AND EMPIRICAL DATA

bodies in the supralunary realm. The two key contributing aspects of comets are: (a) a fiery principle resulting from upward motion in the realm above the earth (the principle of ignition) and (b) a suitably condensed mass of the hot dry exhalation (the fuel). Aristotle further remarks that the phenomena resulting from combinations of these two causal components may vary in shape. They can take either the more familiar from of a long-tailed comet or instead the shape of what "is called" a bearded star (kaleila p6g6nias; 344a23). This is an important passage, for it provides an example of how scientific understanding may lead us to group together two phenomena we had formerly described by different names, since they turn out to have the same essence.12 Aristotle uses his own account to explain the various facts about comets he set out in his earlier critique of the endoxa: The reason why comets are few in number and infrequent, and why they appear more outside the tropics than within them, is that the movement of the sun and stars not only separates off the hot substance but also disintegrates it as it is forming. (345a6-9) Since Aristotle believes that sometimes the appropriate masses of hot dry exhalation occur in the vicinity of a star, he can also account for why a comet may, but need not, occur in conjunction with a star: When therefore the material gathers in the lower region, the comet is an independent phenomenon. But when the exhalation is formed by the movement of one of the stars - either of the planets or of the fixed stars then one of them becomes a comet. (344a34-bl) Comets do not necessarily appear in any particular part of the sky because the hot dry exhalation could be massed anywhere. Several comets might appear in certain rare cases in which the comet-making components occur at more than one place in the sky. And so on.· In much the same way, Aristotle's proposed explanation of hail is couched in terms of his hypothesized fundamental exhalations. He does, however, rely upon an additional complex principle .(which he takes to have been endorsed previously): the basic idea is that heat and cold can "condense" one another. In other words, heat applied to something cold will, at least initially, compress the coldness inward, intensifying it. Once more, Aristotle's account of the phenomenon he is studying, hail, has formal and material aspects. In warm seasons, he explains, the heat outside a cloud can condense the coolness in that cloud when it descends into the warmer regions near the earth. Then

the cloud's internal moist exhalation is rapidly frozen; since this makes it heavy, it falls to the earth as hail. On this account it is possible, Aristotle believes, to explain some of the puzzling facts we initially observed about hail. For instance, Hail is rarer in the summer than in spring or autumn, though commoner than in the winter, because in summer the air is drier; but in spring it is still moist, in autumn it is beginning to become so. (348b26-29) Again, it hails less in high places because the air surrounding clouds is cooler there. And further, larger hailstones come from clouds closest to the earth because such clouds have been more subject to the "heat­ compresses-cold" principle.

·

-

.

'

:1 !i i

11

iii

297

C. A. FREELAND

(12} In other words, the nominal definition of comet as "long-haired star" directs scientific attention toward a certain subject though it incorrectly indicates the essence of that subject. See Bolton (1976}.

G. ENDOXA AND SCIENTIFIC METHOD

·

What conclusions can be drawn about the role of endoxa in science based upon examples from the Meteorology? It may seem simply that Aristotle gains a kind of rhetorical advantage from his method of presenting, and then summarily dismissing, his predecessors' best efforts. This, however, is too simple an assessment. As I see it, Aristotle uses his surveys of prior theories to accomplish two related tasks. First, in those cases in which he includes such surveys, i.e. precisely when studying the most difficult, intriguing phenomena of this science, he presents the basic data or empirical facts about each phenomenon within the context of his refutation of an earlier view. What may at first seem to be merely a matter of convenience, or perhaps a rhetorical move aimed at general one-upsmanship, might instead indicate a special way in which Aristotle sees science as a sort of cumulative group endeavor. That is, he regards it as important to see where an existing scientific theory fails, because it makes a false prediction about something that one might otherwise nol have considered relevant, so might not otherwise have observed. An account of comets as planets, for example, predicts that they should appear only within the zodiac circle; so we are made aware that location in the sky is an important feature of comets and we acquire an important directive for seeking data which go beyond the initial observations about comets' most noticeable features, their t.a ils. Or, if a theory of hail such as Anaxagoras, indicates that larger hailstones should fall from higher clouds, then as scientific observers we are prompted to gather certain phainomena about hail and perhaps to notice heretofore unobserved details about what kind of hail falls from what kind of cloud, and in what location and season. In their second key role, Aristotle's surveys of endoxa reflect a picture of science as a problem-solving activity. Aristotle's critical discussion of an earlier scientific explanation, such as Anaxagoras'

.' 11

I:I I

I: I i

I

i

I

I

'

'I

SCIENTIFIC EXPLANATION AND EMPIRICAL DATA

account of hail, does not merely direct him and his reader/listeners toward an accumulated list of data about hail, it also creates a puzzle or focuses a question about the phainomena. A key question about hail is not just what it is (or, in the rephrased version we might expect frorh certain passages of the Posterior Analytics, why does ice fall from the sky?), but rather, why does ice fall from the sky less often in winter than in fall or spring? After all, it's colder in the winter, and ice is formed by cold. Or in other words, why does hail behave differently than snow? The complexity of the scientific inquiries in the Meteorology is much greater than that allowed for by the Posterior Analytics. Perhaps Aristotle's discussion of scientific definition in the Posterior A nalytics, particularly in Book II, 8 and 10, is meant to be very schematic. When he asks about what thunder or eclipses are, he suggests that the scientist will look for a cause which can be incorporated into an appropriately grounded definition. For example, he suggests, thunder might be defined as the noise of fire being quenched in the clouds. This is certainly not his own account, and furthermore, in an actual scientific inquiry about thunder he will raise more specific questions, e.g.: Why does lightning only sometimes occur with thunder? Why is thunder associated with certain types of clouds? What distinguishes thunder from other loud noises in the sky? These sorts of questions arise because the inquirer has paid attention not only to the empirical phainomena but also to the endoxa. Of course, the Posterior Analytics emphasizes the need for explanatory complexity when it a-d mits that there are multiple (viz: four) possible types of causal explanation. It is helpful to juxtapose Aristotle's view of scientific explanation to more modern ones, such as that of Bas C. van Fraassen's in his book The Scientific lmage. 13 Like Aristotle, van Fraassen is concerned about the assymmetries of explanation.14 In partial response he reviews recent analyses of the logic of questions, including those of Belnap and others.15 Focusing particularly on why-questions, Van Fraassen suggests that they may be differentiated according to topic, contrast-class, and relations , of explana­ tory relevance (pp. 1 4 1 ff). It is not necessary to go into the details of van Fraassen's exposition of these aspects of why-questions in order to recognize their usefulness for describing Aristotle's approach to the endoxa in the Meteorology. A scientific inquiry is not only launched but

aclually gets its identity from the specific questions it asks. The Meteorology shows that Aristotle finds reviews of previous scientific theories crucial to the process of refining why-questions. As I have suggested above, for example, through his survey of the endoxa Aristotle progressively refines the topic of his why-questions about comets. The focus is no longer just on "Why are some stars long-haired?" but "Why are some long-haired stars so different in their behavior from other stars?" The contrast-class for comels clearly includes planets, since when the scientist asks "Why do comets appear anywhere in the sky? , " there is an implied contrast to planets, which are restricted to the zodiac. And again, questions about the endoxa help to indicate just what will be explanatorily releuant. In relation to comets "Why are there numerous comets now?" involves a particular emphasis: "Why now, during a hot and dry season , rather than earlier, during a rainy cool season?" By surveying endoxa the scientist has been directed to observe which seasons are relevant to the appearances of comets. Of course it could be objected that an Aristotelian scientist does not need to survey previous theories either to supply an initial list of phainomena or to order to acquire a clearly focused list of why­ questions. Such data could be gathered and questions could be formulated de novo. I am unclear about how Aristotle would respond to this point. Comets are rare phenomena, not easily observed. Is it merely helpful to prepare for a possible viewing of a comet by reviewing earlier scientific accounts, or is it essential? My general sense is that, despite occasional passages suggesting otherwise (e.g. Pr. An. I, 30), Aristotle is nol committed to an extreme form of (Baconian) empiricism, according to which an individual scientist directly confronts and draws hypotheses from the great masses of sensory data available (concerning comets, hail, or whatever) through the use of simple induction alone.16 It would be equally wrong to attribute to him the Popperian thesis that the goal of science is gradual progress via a slow process of accumulating data which will directly falsify earlier theories - far from il.17 In Aristotle's view, science can at least in principle be "finished,"

(13) Van Fraassen (1980 b). In their 1988 paper Kakkuri-Knuuttila and Martin Kusch describe a general "question-theoretic" approach to philosophy of science as linked to both the Aristotelian tradition and the hermeneutics of Collingwood and Gadamer, and with such modern proponents as Nicholas Rescher (Rescher 1982), Jaakko Hintikka (Hintikka 1976, 1979) and M. Sintonen (Sintonen 1984). (14) See also Van Fraassen (1977) and (1980a). (15) Belnap (1963).

i

!I

1

II

299

C. A. FREELAND

298

(16) This is a very difficult topic which I cannot address in detail here, as it would lead too far afield from the lvfeteorology. See, for some discussion of how Aristotle's description of the intuition of first-principles in science fits with his overall empiricism, Barnes (1975), pp. 248-260, esp. pp. 259-60. Also relevant are discussions in Owen (1961), Nussbaum (1982), and Bolton (1988) about the extent to which Aristotelian examinations of phainomena include conceptual analysis. Bolton criticizes an interpretation found in Barnes (1 980) and Irwin (1981); as I noted above, Brunschwig (1987) suggesLs it is irnportant that divergent opinions are often grounded in studies focused on different portions of the Aristotelian corpus. (17) As expounded in Popper (1962). Kakkuri-Knuttila and Kusch (1 988) make the point that both the crude empiricist and falsificationist (or Popperian) rely on a simple view about how directly empirical observations can bear on scientific

300

C. A . FREELAND

SCIENTIFIC EXPLANATION AND EMPIRICAL DATA

and (as we shall see below) certain scientific hypotheses can be actually confirmed. What I have been suggesting here is another way to understand his insistence upon learning from previous thinkers. I n particular, I have argued, the scientific investigator examines earlier views in order to refine why-questions and focus the direction of empirical observations. Of course, this should not be taken as grounds to attribute to Aristotle the extreme Kuhnian view that empirical observation in science is always guided by existing th_eoretical paradigms. 18 One clear indicator that Aristotle has no concept of paradigm-guided science is his piecemeal approach to the "theories" of previous thinkers.

Now, meteorology even today seems to be a science concerned with the informed interpretation of signs. We say along the Gulf Coast that muggy weather and soft breezes are a sign that a cold front is about to pass through; in the M idwest people interpret dark green clouds and rapidly decreasing barometric pressure as signs of an approaching tornado; and I have heard it alleged that the behavior of cows or other animals can give some sign of an impending earthquake. Does Aristotle's use of signs in this case amount to anything more than this?'° I think there is no doubt that it does. Aristotle is actually concerned less with treating various meteorological phenomena, in the sense of things, as signs, than with analyzing why they can be read as signs. The terms "sign" (semeion) and "proof" or "index" (lekmerion) are introduced into technical usage in Prior A nalylics I I , 27. In a recent discussion of this chapter, Myles Burnyeat has noted that here Aristotle is more interested in a concept of signs as inferences than in the more ordinary sense in which we might call spots, say, a sign of measles.21 Aristotle explicity defines the term "sign" via its function in certain kinds of reasoning: A sign ... means a demonstrative premiss which is necessary or generally accepted. That which coexists with something else, or before or after who.se happening something else has happened, is a sign of something's havmg happened or being (70a8-10). Aristotle contrasts a sign to a "probability" (eikos), although he may mean (as Burnyeat believes) to treat the latter as a type of sign the type involving "generally accepted" rather than "necessary" premises. Syllogisms, here called enthymemes, may involve either probabilities or signs. Aristotle must mean that reasoning with either is enthymatic or incomplete when only one premise is stated. In general Aristotle holds that "truth can be found in all signs" (alelhes men oun en hapasin huparxei lois semeiois; 70a38) . However, signs differ according to the sorts of syllogistic reconstructions which may be offered to represent the inferences they involve. Particularly valuable are signs representable by syllogisms in the first figure: "[A] syllogism in the first figure cannot be refuted if it is true, since it is universal." To illustrate this sort of case, Aristotle cites this example: Having milk is a sign that a woman is pregnant. There is, supposedly, a universal connection between having milk and pregnancy.22 Logi­ cally, we get this reconstruction in a first figure syllogism:

Part II. Signs and Proofs

A . S I GNS AND PROOFS IN THE PRIOR ANALYTICS So far I have examined Aristotle's use of endoxa in structuring scientific accounts of various meteorological phenomena. In saying that these surveys create a prima facie obligation to provide superior explanations, I have suggested that Aristotle believes that his own accounts surpass rivals in explanatory power. He defends his own proposals not only by alluding to their superior ability to deal with the initially presented list of empirical phainomena, but also by claiming that he can account for certain ne'"'._ phainomena. These new "sumbai­ nonian or accompanying facts are often described as "signs" of various sorts. Aristotle discusses a variety of signs of earthquakes, for example, such as particular cloud-patterns or a special dimmed way the sun might appear on a cloudless day. We now know, thanks to studies of Sedley and others,19 that in the post-Aristotelian scientific tradition there were important debates about signs, also usually regarded as sumbainonld, or accompanying facts. In particular, the Stoics and Epicureans disputed whether such signs played a role in the discovery or evidential support of particular scientific hypotheses. Aristotle's treatment of signs in the Meteorology is a neglected antecedent to this later debate.

11

I! 'I i

hypotheses. Despite Popper's influence, this simple view has been considered dubious in philosophy of science for some time, by Kuhn and others too numerous to list (but including at any rate philosophers as otherwise diverse as Duhem, Feyerabend, GrOnbaum, and Quine). (18) Kuhn (1970). Kakkuri-Knuuttila and Kusch (1988) point out that there are natural links between what they describe as the question-theoretical approach in philosophy of science and Kuhn's notion of paradigm-guided inquiry, and they note that these connections have been pursued by Stephen Toulmin in Toulmin (1972). (19) See Sedley (1982); he discusses, among others, writings by Michael Frede and Gisela Striker.

301

(20) The reader may notice that I sidestep tough questions here about the status of signs in the contemporary "science" of meteorology. (21) Burnyeat (1982), pp. 194-99. (22) Burnyeat discusses the factual problems of this case; see Burnyeat (1982), pp. 204-5, n. 30.

I

'

rJ1 ' u! I ' ·1 !·

• :

' '

!'

I

:

302

C. A . FREELAND

SCIENTIFIC EXPLANATION AND EMPIRICAL DATA

All woman who have milk are pregnant. (All) B is A. This woman has milk. (Some) C is B. This woman is pregnant. (Some) C is A. This sort of case is contrasted to others in which signs may be· represented by invalid second or third figure syllogisms. Aristotle first summarizes by calling a sign in general, as a middle term, an "index" or "proof" ("tekmi!rion") of something: "For the name 'index' (iekmi!rion) is given to that which causes us to know, and the middle term (lo meson) is especially of this nature" (70 b l-4). However, just after this· remark he seems prepared to reserve the label "lekmi!rion" to apply only to those signs which are representable as middle terms in valid first-figure syllogisms.23 In other words, having milk serves as "proof" that a woman is pregnant,24 but other signs may not be proofs. Now it is clearly the case, but still worth mentioning, that a proof is not the "cause" of whatever it is proof of: having milk does not cause being pregnant, but is in fact caused by it. Nevertheless this, like other signs, is "that which causes us to know''. Our awareness of the one fact (that the woman has milk) is grounds for inferring the other fact about the woman (that she is pregnant). Indeed, Aristotle maintains that "the conclusion which is reached through the first figure is most generally accepted and most true (endoxotaion gar kai malista ali!thes)" (70b5-6). But we must be careful here, because strictly speaking, Aristotle holds that only "the" cause, i.e. the true cause, is a cause of our knowing, i.e. validly demonstrating (cl. Post. An. I , 2) the conclusion that the woman is pregnant. The true cause is the basis for proper inference because it represents the order of being as it really is. Aristotle's causal explanation for the presence of milk in humans is provided within his discussion of human pregnancy in Generation of Animals IV, 8. As we would expect, he there describes a final cause-nurturing the infant­ and also a material/efficient causal story. The latter involves .an excess of material in the pregnant woman at around seven months or so. Aristotle thinks that, as the process of baby-building nears completion, uthen there is more surplus residue, because less of it is being used up ... " (776a31-3). This surplus residue takes the form of milk. In other words, pregnancy causes having milk because through insemination the male's heat has introduced a "concocting" agent into the woman's body. This agent directs that her available material be put to service, and seven months or so into the process it also results in her having milk. So, according to the Posterior A nalyiics, genuine

knowledge about "Why is this woman pregnant?" requires knowling the facts about insemination and nurturing infants; and only such know­ ledge counts, strictly speaking, as "the" cause both of our knowing that the woman is pregnant and that (consequently) she has milk. Know­ ledge through signs might show us that the woman is pregnant, but it does not explain why."

This is how Burnyeat takes him, too. See Burnyeat (1 982), p. 196. There are important considerations here about the modal and epistemic status of the true universals involved in syllogistic reconstruction of proofs. Burnyeat discusses these in Burnyeal (1 982), pp. 204-5.

(23) (24)

303

B. SIGNS AND PROOFS IN THE METEOROLOGY I have emphazised these points about various sorts of grounds for knowledge and inference because we need to be clear about them so as to understand Aristotle's claims about signs and proofs m the Meteorology. In an informal survey (based on the citations in Lee's index) I have found 19 occurrences of the term "semeion" together with four occurrences of "lekmerion" in something like the technical (inferential) use set forth in the Prior Analylics.26 It is, to be sure, difficult to force some of these occurrences into any clear inferential pattern; some may be ordinary everyday uses, as when Aristotle speaks of snow as a sign of a cold season or country (!, 1 1 , 347a28). Once he counts something as a sign which corroborates something else he took to be an initial fact: the fact that hailstones are not rounded off is proof that they're frozen close to the earth (I, 12, 347a33). In a more interesting passage about dew and frost he cites the fact that they aren't formed on mountains as a sign that they're due to vapor not rising very far. We will see longer and more elaborate examples along these lines in a moment. One rather remarkable sign mentioned in this text must be counted as among the type Aristotle describes in the Prior Analytics as merely probable or an eikos. This sign is mentioned at II, 1 in his discussion of why the sea cannot have sources, and of how rivers flow from higher parts of the earth, those towards the north. Here Aristotle writes,

An indication that the northerly parts of the earth are high is the opinion of many of the anc.ient meteorologists that the sun does not pass under the earth but round its northerly part, and that it disappears and causes night because the earth is higher towards the north (354a27-32). (25) Burnyeat rejects Ross' ("natural") view that a sign must be more familiar or knowable than what it signifies, and he claims that in the Prior Analylics "Aristotle's ac�ount of sign� makes no use of epistemic notions" (p. 205). According to Burnyeat, Aristotle leaves 1t open that "pregnancy could be a sign of lactation." I agree that this is so. Howev�r, what is crucial about a proof is that it is a cause of our knowing something, _ although on its own 1t cannot count as lhe cause; to obtain the cause we must arm ourselves with a proper scientific demonstration. . (26) A� Burnyeat notes, Aristotle elsewhere uses simeion more informally· to designate things rather than propositions or bases for inferences. See Burnyeat (1982), p. 194, n. 3.

304

C. A. FREELAND

This passage is especially interesting because Aristotle cites as a "sign" (semeion) something which is merely said to have been believed by the ancients-a clear reference to an endoxon which he tries to use to lend support to his own theorizing. The most extended and complex allusions to signs in this work, together with mention of lekmeria, seem to cluster around three especially problematic phenomena in meteorology: comets, earthquakes, and the saltiness of the sea. In the next sections I will consider_the first two of these examples in more detail, to uncover the inferential structure and scientific methodology at work in them.27 c. COMETS

As noted in section IF above, Aristotle uses his own account of comets to explain certain empirical data about comets collected during his survey of the endoxa, such as their rarity, paths in the sky, shapes, and relations to stars. Toward the end of his discussion, and a fter providing his own explanation, Aristotle attempts to lend persuasiveness to his account of comets by discussing a new empirical point: comets are likely to occur during seasons that are windy and dry (344b28ff). He says that a multiplicity of comets "signify" (semainousi) wind and drought (there are subtle variations, such as the claim that when comets are smaller or dimmer, tbe wind or drought are less noticeable). We can draw an instructive parallel between this example of a sign ("Frequent comets signify wind and drought") and the example from the Prior Analylics ("Milk signifies that a woman is pregnant") as follows: All B is A. All seasons with frequent comets are windy and dry. A certain winter was a season with frequent comets. This C is B,. This C is B, A certain winter was windy and dry. Recall that, although milk is a sign of pregnancy, it does not cause it; rather, certain facts associated with pregnancy cause a woman to have milk. So also Aristotle believes that comets signify, but do not ' cause, wind and drought; rather, factors involved in the wind and drought (i.e., a predominance of masses of hot dry exhalation) also cause the occurrence of multiple comets. Each sign could be classified as a "proof" of what it signifies, since in each case the sign is a "most reputable" cause of our knowing the associated phenomenon. It might be objected that there is insufficient evidence to treat the example about comets just discussed as a "proof" or sign representable in a valid first-figure syllogism. Signs which are to count as lekmeria (27) I discuss the third and longest of these cases in Appendix I, "The Saltiness or the Sea".

l

SCIENTIFIC EXPLANATION AND EMPIRICAL DATA

305

could be described as sufficient conditions of what they signify. Aristotle does in fact maintain that the appearance of multiple comets is a sufficient condition of a windy and dry season. Indeed, he argues not only that multiple comets are a sign of wind and drought but that lhis facl (this signifying relation) is ilself a proof (lekmerion) of the correctness of his account of comets-it confirms his view that their constitution is fiery (344bl9-21). Recall that milk is a proof of pregnancy because these two palhe can be connected in a valid first figure syllogism. But this isn't the whole story: these palhe are connected because the actual factors that cause pregnancy (those demonstrable by a scientific syllogism) also explain why a woman has milk. Now here, similarly, multiple comets are proof of a dry windy season because there is an appropriate and universal connection between comets and windy dry seasons. Again, this isn't the whole story: the true scientific cause for windy dry seasons, predominant masses of the hot dry exhalation, would also be cited in the scientific explanation of the frequent occurrence of comets. Here Aristotle goes beyond the Prior Analylics account, by citing a particular sign relation as itself evidence ("proof") for his background scientific explanation; I shall have further comments about this special use of ((tekmerion" in my Section E below. D. EARTHQUAKES Aristotle holds that earthquakes occur when the hot dry exhalation in its dynamic form as wind is forcibly and rapidly thrust into various hollows of the earth. As usual, he collects various data about earthquakes in the course of his preliminary survey of the endoxa (here in I I , 7, he reviews proposals from Anaximenes, Democritus and Anaxagoras). It is especially noteworthy that he criticizes Anaxagoras' view for failing "to account for any of the peculiar features of earthquakes, which do not occur in any district or at any time indiscriminately" (365a34-6). In the very long chapter I I , 8 Aristotle first gives his own account of earthquakes and then proceeds to show how his theory proves superior to that of Anaxagoras in explanatory power. Indeed, he marshalls together an amazing number of features of earthquakes which (he thinks) he can account for; none of these empirical facts28 were noted in his treatment of Anaxagoras or ' indeed ' before he provided his own definition. Thus he explains in considerable detail why most earthquakes occur in calm weather, at night, in places "where the sea is full of currents or the earth is porous and hollow ' ' ,

(28) I t should be obvious that I have been using the term "explain" throughout this paper, but especially in this section, in the sense of "purport to explain," and here similarly I speak ? f "empirical facts" as "what Aristotle takes to be empirical facts".

1,: .

1

i

I

�

I •:"

I

1' I ,

·I .' '.

."' ·.

. :I '

. ',

•

I,(1, I:1 i!: ' 1:

1

.

"

SCIENTIFIC EXPLANATION AND EMPIRICAL DATA

most often is spring and autumn or during rains and droughts, at an eclipse of the moon, with aftershocks, with noises preceding them, with accompanying jets of water or tidal waves, in narrowly circumscribed localities, in the form o f either (horizontal) shuddders or (up-and-down) throbs, more rarely on islands far out to sea . . . , etc. And he concludes with perhaps j ustifiable pride by saying "This completes our examina­ tion of the nature and causes of earthquakes, and of their most important attendant circumstances (sumbainonla} (369a8-9)" . Now, i n the course o f setting forth this whole series o f explanations Aristotle several times mentions signs or proofs. In the first passage he claims that there are "signs of these things" (viz: of his account) in certain observations about earthquakes which didn't cease until wind broke out with hurricane force (366b3lff). Again, we can call upon the Prior Analytics discussion to represent the inference Aristotle has in mind as taking a form something like this: In every case of turbulence where the wind escapes, the earthquake ends. Jn this case of turbulence the wind escapes. In this case of turbulence the earthquake ends. Escaping wind is a sign of an earthquake's ending since an earthquakes is a sufficient condition for trapped wind. Once more, the sign reveals the phenomenon because it indicates something that is causally connected to it. In the next paragraph, Aristotle speaks of a proof (tekmi!rion) that winds circulate beneath the earth: For when a south wind is going to blow it is heralded by noises from the places from which eruptions occur. This is. because �he sea, � hich is being driven forward from far off, thrusts the wind that is erupting out of the earth back again when it meets it. This causes a noise but no earthquake because there is plenty of room for the wind, of which there is only a small quantity and which can overflow into the void outside. (367al5-20) ·. This passage claims first that noises within the earth are a sign of a (south) wind. It then explains that this signifying relation holds because Aristotle's account of earthquakes is correct: the rumbling noises signify south wind because they are like a mild earthquake which occurs due to the wind's being trapped under the earth. Here and in the next use of the sign terminology in this chapter Aristotle is arguing, just as he did concerning comets, that his ability to explain particular signs indicates that his accound is correct. He notes, Further evidence that our account of the cause of earthquakes is correct is afforded by the facts that before them the sun becomes misty and dimmer though there is no cloud; and before earthquakes that occur at dawn there is often a calm and a hard frost.

conditions for the presence of the crucial causal factors of earthquakes. That is, there is a universal connection between the misty dimmed sun on cloudless days and earthquakes; in his next paragraph Aristotle uses the phrase "necessarily" three times and also speaks about causal processes which follow from the essential nature of heat by itself (kath'hauli!n}. A misty dimmed appearance of the sun on a cloudless day signifies an earthquake because both are caused by the same general factors: "The sun is necessarily misty and dim when the wind which dissolves and breaks up the air begins to retreat into the earth " . Cold and calm also signify earthquakes because they too are related to the "right" causal conditions: Before an earthquake, "the exhalation, which is by nature essentially warm, is directed inwards": So the warm element disappears into the earth, and wherever this happens, the vaporous exhalation being moist condenses and causes cold. Yet another closely related "sign" of earthquakes can be similarly explained. Sometimes earthquakes are signalled by a fine long streak of clouds in the sky or of breakers in the sea: The wind produces the same effects on the cloud in the sky as the sea on the shore, so that when there is a calm1 the clouds that are left are all straight and fine like breakers in the air. Here too, the sign that indicates an earthquake does so, Aristotle tell us, because it is related to what he has argued is the real cause of earthquakes, the exhalation (suitably qualified). Aristotle explains numerous other features associated with earthquakes in terms of his core account (of exhalation rushing inwards) ; though they are not explicitly referred to as signs, it seems likely that in many cases the pathi! under consideration could count as such. Thus the noises in the earth before earthquakes could be heralds just as the fine line of clouds would be; or, again, the fact that Aristotle can explain aftershocks should be taken to lend credibility to-or to serve as a sign for the truth of-his causal accounts. In one final case which is importantly different Aristotle cites something which could not be treated as a sign of earthquakes in this same sense, namely, an eclipse. He remarks that "sometimes" the two phenomena are associated. Perhaps, since the connection between these two phenomena is not universal, an eclipse is only a somewhat trustworthy sign of an earthquake. Even in this case there is some background cause linking the two phenomena: a type of wind that was being held in counterbalance by heat from the moon rushes back into the earth as the moon's heat subsides during the eclipse. Thus an eclipse is a fter all a sign, a somewhat trustworthy bit of evidence that something is going on which will probably affect balances of light and warmth affecting the earth's wind systems; and these imbalances in turn might get translated into the dramatic causal forces resulting in an earthquake.

·

I

I•

'

i I

! ;

In this passage Aristotle mentions two concomitants (sumbainonla) of earthquakes; these signs (si!meia) can again be regarded as sufficient

I' i

'

Ii A�

307

C. A. FREELAND

306

.l

T

I I

.

J

. ; .,i

308

C. A. FREELAND

,I

E. SIGNS, PROOFS AND ABDUCTION In these arguments of the Meleorology Aristotle is doing far more than informing readers about how to interpret various weather signs; he is employing a venerable, if questionable, strategy for the defense of scientific realism ' the strategy of "abductio n" . Roughly, this strategy involves arguing from explanatory success to the . truth of a s<:ientific theory. As a scientific realist Aristotle is committed to mamtammg that it is in principle possible for a scientific theory to provide true and accurate accounts of the actual causes of the empirical phenomena it investigates. In meteorology, for example, Aristotle would maintain that the exhalations which are fundamental principles of meteorology really do exist and do function in just the ways his theory describes." What evidence can be cited for this conviction? Abduction is a form of inference to the best explanation: the best explanation for a theory's predictive success is that it is true, i.e. that it describes the world as it really is. Indeed, abductive arguments maintain. that "the (approxima­ te) truth of a scientific theory is the only possible explanation of its predictive success" .30 Even if Aristotle is arguing merel.y that his theoretical explanations seem workable, and that this constitutes some reason to believe them, he is employing an interesting and controversial strategy.31 But he often (either explicitly or implicity) argues for more, that his theory's workability provides a proof of its correctness, or a valid ground for the inference that his causal story is a true on.e. Contemporary anti-realists criticize abductive arguments as circular or question-begging" because abduction involves a sort of defense of induction through the use of induction. In a recent discussion, for example, Richard Boyd briefly sets out the antirealist argument as follows: The issue of scientific realism is-at least insofar as the dispute between realists and empiricists is concerned-a debate over the legitim�cy of inductive inference to the best explanation ... Arguments for realism

\:

I

I

{29) However, Aristotle's chapter on comets contains a puzzling and anti-rea list _ methodological remark which I discuss in Appendix II, "An Anomalous Methodolog1cal Remark". (30) As phrnsed by Leplin (1 984), p. I . . .. . . (31) For Epicurean uses of something like this strategy - 1.e., c1t1ng compallb1hty with sumbainonla as a form of evidence for a theory - see Sedley ( 1982), esp. pp. 266 ff. on non-contestation. See also Appendix I I below. (32) However, this topic is most often addressed in relation to scientific theories referring to theoretical entities, so what is at stake in particular are realist commitments to the existence of these entities. For a helpful introduction, see Leplin (1984), esp. pp. 17. Also relevant are the essays included in Churchland and Hooker (1985).

lrII 1I•·;:

.

1 ,I

1 ·!'

ii J ,

:I

I

'i\ !

,u

SCIENTl·FIC EXPLANATION AND EMPIRICAL DATA

309

employ just this sort of inference, and thus, simply beg the question between realists and empiricist antirealists.33 How exactly could Aristotle defend his abductive strategy of citing signs as evidence for the truth of his own causal hypotheses? In some cases he could point to background deductive arguments; .as I have indicated above, he views certain especially interesting signs as sufficient conditions of what they signify. In these cases, reasoning from a sign to a phenomenon is valid because, as Aristotle explained in the Prior Analylics, there is some universal connection between relevant palhe (e.g. pregnancy and milk) which can be represented in a deductively valid first-figure syllogism. But what provides such syllogisms themsel­ ves with scientific support? How, for example, can the scientist be sure of possessing the right background syllogism of a science, for example a demonstrative explanation from biology which would explain why women get pregnant? Here Aristotle is arguing that usefulness provides at least partial corroboration of such a causal account. There is more than one reason to believe one has the genuine scientific account if the same causal story can explain several palhe. Perhaps Aristotle's treatment of signs as inductive evidence for the 1 truth of scientific claims reflects his more relaxed use of "sign' in Prior Analylics I I , 27; recall that there he maintained that signs of all types have a role in revealing truth. Signs point us toward an understanding of the real cause of a signified phenomenon. Burnyeat emphasizes in his review of the Prior Analytics doctrine of signs that ... the purpose of this technical exercise is not to reject the inferences which do not admit of a formally valid reconstruction (Burnyeat 1982, p. 196). As Burnyeat sees it, in allowing all signs some value in indicating truth, Aristotle opens the door to forms of non-deductive inference. This move is crucial for the philosophy of science; indeed, Burnyeat believes that Aristotle's logic is more open and potentially 'useful for philosophy of science than that of the Stoics who followed him: If one believes that an adequate philosophy of science must find a place for nondeductive as well as for deductive logic, one will conclude that, as logicians, Aristotle was a better friend to the sciences than Zeno and Chrysippus (Burnyeat 1982, p. 238)34

(33) Boyd (1984), p. 66. The anti-realist he is describing is Arthur Fine, and the realist arguments mentioned here are said to be those "of the sort Fine criticizes". (34) For further discussion of development of signs within philosophy of science, both by the Stoics and by Epicurus, see also Sedley {1982). The role of signs in confirmation in the Meteorology, as I proceed to interpret it, has interesting parallels to their role in Epicurus' theory, as set forth by Sedley; see especially his pp. 267-69 and 27071.

310

i i' 1 ,' j

'i I

' I' '"' I1 l i

i ; ' I :. ,

I;

\i

i·! 11 1 :1

SCIENT-IFIC EXPLANATION AND EMPIRICAL DATA

C. A. FREELAND

really invoking a sort of me/a-sign. That is, he takes the fact that something is a sign as itself the sign of something else-of the truth of his analysis of comets. This move is not exactly envisaged in the Prior Analytics discussion of signs, but neither is it incompatible with that discussion. Aristotle seems to argue as follows: The fact that comets are a sign of wind and drought is a sign that my account of comets is correct. '!' he possessor of a true and adequate account of causes can also explain why comets are a sign of wind and drought. We can compare this reasoning to a parallel reconslruction of his Prior Analylics exposition:

In saying that a sign supplies evidence that is endoxalafon or malisla alelhes (most reputable and most true) for a conclusion, Aristotle "implies that the inferences reconstructed in other figures do have some, though a lesser, claim to reputability and truth" (p. 197). Even non­ "proof" signs represent "certain common and useful inferences of ordinary life" (p. 1 96). I believe that Aristotle accords certain signs an important evidential role in arguing for the truth of his proposed definitions of meteorological phenomena. Even a sign of a non-proof category can serve as a "reputable" (endoxon) indicator of a fact. This is so because all signs are related in some way to a properly demonstrative syllogism. An eclipse, for example, is a sign but not a proof of an earthquake." In this case Aristotle offers a single causal explanation which helps account for both phenomena, without requiring that they be universally or necessarily associated. The sign is a sign because there is a background link ; furthermore, Aristotle cites the sign as partial evidence that he is right about this link. In other cases the signs Aristotle describes have closer and more integral connections to the phenomena they signify, even if they are not explicitly regarded as "proofs". Thus Aristotle remarks that a misty dimmed sun signifies an earthquake; and he also explains why. He argues that this sort of sun is a sign of earthquakes because his causal account of earthquakes is true. What is the structure and purpose of this argument? Aristotle marshalls explanations of why various things as well as this one are signs of earthquakes in order to defend the plausibility of his own account of earthquakes. He believes that he can go beyond explaining the initial data about earthquakes, so as to explain the sumbainonta, the associated empirical facts which are signs of earthquakes. Thus signs function evidentially in support of his c�usal hypotheses. As Burnyeat suggests, signs allow Aristotle the use of certain everyday forms of inference. In the Meteorology Aristotle uses signs as part of an argument form involving more than inductive inference lo phenomena (of the sort practiced by everyday "weathermen"); instead, using an abductive strategy, he constructs inferences to the correctness of his accounts of phenomena. The structure of such inferences is very clear in those cases in which Aristotle describes certain sign relations as themselves signs of the accuracy of his own accounts. When Aristotle says that the fact that comets signify wind and drought is itself a sign (tekmerion) of the correctness of his causal analysis of comets, he is

The fact that having milk is a sign of pregnancy is proof that my account of why wo�en have milk is correct. There must be a true and adequate explanation for why women have milk, and the possessor of this true and adequate explanation can also explain why milk is a sign of pregnancy. Aristotle reasons in this complex way in the Meteorology both here in his claim about comets and also in the passages I discussed about earthquakes. Certain signs are valid indicators of phenomena because they bear a special relation to the true causal analysis of those ph � nomena; and Aristotle would cite this same point in explicating the . Prwr A nalytics example of a tekmerion. It is just this italicized "because" that Aristotle invokes with his use of meta-signs in the Meleorology. Other signs function in less rigid argument-patterns as indicators that Aristotle's definitions of meteorological phenomena are correct. These signs usually involve empirically observable facts phainomena or sumbainonla. These facts are functionally parallel to th� facts cited earlier during Aristotle's surveys of endoxa; they can be paired up with the initial endoxa in three significant ways. First, all these facts could be cited in a new and revised list of the endoxa, or most reputable facts of this science. Second, there are all particular sorts of endoxa, empirical facts available to aisthi!sis; as such, they ought to be accounted for b y the scientist even if they are not universally or . . necessarily associated with the subject under investigation. Third, and finally, by claiming that his hypotheses do explain all these facts, both those recorded initially and those added on as signs or proofs, Aristotle provides inductive evidence for the truth of his own scientific theory. Part III. The Pragmatics of Explanation

How is Aristotle's procedure in the Meteorology related to the formal outline of scientific method he presents in the Posterior Analytics? On the most obvious level the Meteorology differs from the pure model by its inclusion of the two main features I have examined in

{35) Surely a variety of the phenomena Aristotle calls signs of earthquakes will also fall into this category of non-proof signs; see also Appendix I below, on the saltiness of the sea.

1l

l' II

, 'I I '

d,

31 1

'I

312

)

l

i

I ; I

'I

! i

'

,',! i

h

'

313

develops between "knowledge" and "understanding" is said to reflect "two things which present day philosophy segregates into distinct areas of inquiry (p. 97)". These are first, "an account of the conditions for a proposition to belong to a body of systematic knowledge like geometry, physics or botany" ; such an account belongs, says Burnyeat, to the d1sc1phne of p hilosophy of science. Science, there is "an account of the cognitive state of the individual person who has mastered a body of . systematic knowledge"; this second thing belongs to the realm of epistemology. " Knowledge" is taken to be roughly sketched out as in traditional epistemology which views it as justified true belief. Scientific knowledge or "understanding" requires incorporating a known item m specified ways within a larger body of information with unifying laws and principles. Here Burnyeat cites Michael Friedman 's article, "Explanation and Scientific Understanding"" as setting out the paradigm (and he thinks it one Aristotle could agree to). As Friedman has characterized it, "scientific explanation is in the first instance explanation of generalities (laws) rather than the explanation of particular events" (p. 109). But as a matter of fact, explanation is a very controversial notion in contemporary philosophy of science. In a recent article "Explanation and Realism" , Clark Glymour notes that there are three main contemporary approaches to scientific explanation: Philosophical theories o � scientifi � expla �ation can be roughly divided into three types : purely .logical theories, which analyze explanation solely in terms of log1c� l r�lat1ons an ? tr�th conditions; theories with extra objective structure, wh1c� impose obJective conditions on explanation beyond those of �rut� and logical structu_re; and theories with extra subjective structure, which impose on explanations psychological conditions of belief, interest, and so forth.40 Glymour places Friedman, whose account Burnyeat relies upon, in . his seco� d category. F? r our purposes, however, it is more interesting to consider representatives from Glymour's third category n Peter Achinstein, for example, in The Nature of Explanation criticizes certain philosophers like Carl Hempel for regarding explanations as entities of certain sorts rather than focusing on explanatory acls.42 On his view, a

this article, surveys of endoxa and citation of signs and proofs.36 But this question is made more difficult to answer by recent disputes about the actual purpose of the Posterior A nalytics' outline of demonstratively rendered scientific knowledge. In his article "Aristotle on Understan­ ding Knowledge", Myles Burnyeat has challenged the interpretation which he says "promises to become a new orthodoxy ", that advanced in Jonathan Barnes' Clarendon edition of the Posterior Analytics.37 In brief, Barnes argued that the Posterior Analytics advocates demonstra­ tion not as a method of discovery or a model of explanation, but as the best organization for teaching or imparting knowledge. Burnyeat responds with two critical points. First, he argues, Aristotle was himself too good a pedagogue to have believed that demonstration from first principles was the most effective way to teach or convey knowledge (pp. 1 1 6-17). I nstead, he frequently emphasizes the need to begin by citing what is more familiar "to us" or to the students of a discipline. (This way involve beginning from the endoxa, for instance.) Second, Burnyeat presents an elaborate case for translating Aristotle's term "episteme" (usually translated as "scientific knowled­ ge") as "understanding". His point is that, for Aristotle, episteme involves not only knowing a proposition but being in a position to first from it of demonstration suitable by it explain principles. Burnyeat emphasized that a proposition can be known without being understood-indeed, this is likely to be the position of a young student or learner. Converting knowledge into understanding may ultimately involve conveying the science's first principles to the learner; but more immediately what is required are some first steps which enable the learner to begin making the proper connections to other familiar items. Burnyeat's interpretation of Aristotelian episteme as "understand­ ing" is problematic for a variety of reasons.38 The key contrast he

' !.

'

SCIENTIFIC EXPLANATION AND EMPIRICAL DATA

C. A. FREELAND

{36) This is leaving aside questions about just how Aristotle's de,finitions in this science conform to the demonstrative/syllogistic guidelines set out in the Post. An. In Freeland (1986) I argue that there is general conformity, although Aristotle's use of multiple causal accounts (e.g. accounts combining both material and formal causes) of the same phenomena complicates the picture. (37) Burnyeat (1981); Barnes (1975). (38) On Burnyeat's view, a single proposition can be the object of two distinct cognitive states. At first it is known and subsequently it is understood when the knower acquires a demonstration or it. But Aristotle describes such shifts in cognitive slate as, indifferently, shifts in the objecl known. For example, he writes in the Poslerior Analyfics I I, 8 that a scientist who knows the definition of either thunder or an eclipse will know the· relevant syllogism. Knowing the essential nature of either phenomenon is just knowing its cause: that thunder, for example, is "the quenching of fire in cloud" (he says here). Notice that it makes no sense to say that one could "know" this proposition in Burnyeat's sense of "restricted knowledge": to know it is to understand it.

(39) Fc;edman (1974). (40) Glymouc (1984), p, 178. (41) Glymour mentions Achinstein, Skyrms, van Fraassen, and ("recently") Putnam. (42) Achinstein (1983); see esp. pp. 5-7 and Chapter 3, "What is an Explanation?" Actually Achinstein criticizes Aristotle along with Hempel, but I think his understanding of Aristotle is limited, and at any rate he only mentions Aristotle briefly. I t may be that Aristotle holds the ontological view of explanations that Achinstein attributes to him (and others, like Hempel) although he is still sensitive to the sort of pragmatic concerns emphasized by Achinstein and others. I will discuss these topics further in Freeland (1991).

I

I ' '

;i I

!1 ' • ;

I

;i

Ii

i

:J I '

'' , I ::

'�

314

C. A. FREELAND

take into theory of explan atory acts is more impor tant because it can sis. empha and force onary illoculi of issues account the important en, Fraass van Bas is ry catego third this of ntative Another represe g 4 writin In whose book The Scienlific Image I have mentioned earlier. 3 credits ly actual about the "pragmatics of explanation" Van Fraassen the sort of Aristotle's multiple-causal analysis as incorporating just 44 draws an also en considerations Achinstein emphasizes. Van Fraass roles alory explan and tive important distinction between the descrip why­ single a be to seems played by the statements of a science. What explanatory question might actual ly involve a variety of possible answer to a the be might ent statem tive concerns. Or again, one descrip ent may statem same the fact, In s. variety of explanation-question context: upon ing depend roles, atory explan play both descriptive and he format . re t)ion ...[I]f you ask a scientist to explain something to you, thek in fro d1ffe lo o sound ?1 not � does (and ? � . gives you is not different in kind Sn�1la�ly 1� the information he gives you when you ask for a descr1pt10� . the rise in 011 'ordinary' explanations: the information I adduce to explainof request s for prices, is information I would have given you to a batterJ'. an call To ption. consum oil and rs, produce oil s, supplie oil of tion descrip of sort he or form its abou � nothing say � explanation scientific, is to to get information adduce d, but only that the explanation draws on science the this information (at least to some extent) and, more importantly, that good an explanation it is, are being applied criteria of evaluation of how using a scientific theory . ..45 lar Because explanation is a description which answers a particu . twoa simply not as ation why-question, Van Fraassen describes explan n relatio erm three-t a rather but term relation between theory and fact, en between theory, fact, and conlexl. In describing context Van Fraass my in ned mentio (also estions why-qu of further discusses the logic and section IG above). In particular, the features of "relevance" , an from sought really is what ining determ to crucial "contrast-class" are explanation. a­ This extremely minima l sketch of Van Fraassen's view of explan at's Burnye tion should still be enough to suggest a viable alternative to at interpretation of how Aristotle views scientific explanation.' Burnye rm two-te simple represents Aristotelian understanding as a fairly of the relation between a proposition or fact and the first principles in the t interes le's Aristot theory it is embedded in/deriv able from. But should s account causal e assymmetries of explanation and in multipl x alert us to the fact that explanation is for him a more comple which estions why-qu answer to dness relationship, one involving prepare

(43) Van Fraassen (1980b). (44) Van Fraassen {1977, 1 980 a). (45) Van Fraa8'en (1980b). pp. 155-6.

l

SCIENTIFIC EXPLANATION AND EMPIRICAL DATA

315

arise for observers and in coniexis.46 As I argued above, Aristotle uses evaluations of the endoxa in this science to refine why-questions; this creates an expectation about what it will take to "understand " say ' comets or hail. This scientist who truly understands comets has i � hand a descriptive theory involving the hot dry exhalation, but this involves "knowing" more than a mere definition, and it also involves "understanding" more than simply how to deduce this definition from certain better known first principles. In virtue of knowing that "a comet is a combination of the hot dry exhalation with an igniting principle," the scientist also knows the answers to a host of why­ questions, so "understands" and is prepared to explain why comets are ra re, why they appear anywhere in the sky, sometimes but not always . with a star, more often in dry season, etc. Similarly, the person who "understands" hail knows not not only how to define hail by citing first principles - the exhalations - but also can explain the answers to a variety of why-questions about hail: why does bigger hail fall from clouds nearer the earth, why doesn't it hail in winter as much as in the fall, why do hailstones get rounded under certain conditions, etc. And the person who understands earthquakes can explain why they . sometimes occur during eclipses of the moon, or are heralded by a dimmed and misty sun. If this context-dependent account of explanation indeed belongs to . Aristotle, then we must recognize a second major limitation of Burnyeat's interpretation. Not only does he draw a sharp distinction between knowledge and understanding, but he also distinguishes processes of understanding and of scientific discovery. Notice how these two are dichotomized in the following passage from Burnyeat's article: Teaching in the sense of imparting knowledge to people who did not have it before must normally i�clude the citing of evidence and justification. The . led to knowledge which is new to him cannot be _ the pupil is path by which wholly unconnected with the path by which the teacher won that kno�ledge in . the first place. (I am referring here to the evidential base for _ discovery, not to the methods used in the search.)47 a sc1ent1fic

(46) This may seem puzzling in light of Achinstein's criticism of Aristotle for his emphasis on entities rather than illocutionary explanatory acts. I believe that van Fraassen. has better understood the impact of Aristotle's use of multiple forms of explan��1on; see Van Fraassen . (1 9�0 a), . "A re-examination of Aristotle's Philosophy of Science . But I am also arguing 1n this paper that Aristotle places more emphasis on "the pragmatics of . expla.nation" in an actual treatise like the Meteorology than in his formal model for science in the Posterior Analylics. I return to the topic of Aristotelian pragmatics of explanation in discussing the Physics on "accidental causes" in Freeland {1991 ), currently in ms. draft. (47) Burnyeat (1981), p. 1 17.

316

,:: 1 1 ·'

I' I

i

'

:I

I !

I;

,I ,[

SCIENTIFIC EXPLANATION AND EMPIRICAL DATA

C. A. FREELAND

te the But as I see it, it does not make good sense to separa does eat Burny as ," search the in used ods "meth "evidential base" from a for base in this passage.•• It is certai nly true tha.t the evidential first­ ble onstra scientific definition will involve some allus10 n to indem these "self­ er discov to have may er inquir ific scient . principles; the . in the begin evident" principles through induc t10n. (Aristotle does but he does not Meteorology , roughly, by setting out his first-principles, us and self­ obv10 are simply present them with the claim that they _ s endoxa survey n eve and certifying; rather, he argues for . them . . le onto stumb might ist scient concerning them.) Again , a workmg recount o � essar unnec s proces a in � accounts of particular pheno mena, . 0ns efimt1 unts/d acco these for ent . in providing subsequent argum _ ntific . scie to more is there out, d pointe But, as I have just . of defimt10ns understanding and explanation than the mere possess10n a correlative es involv ding" erstan "Und les. derived from first-princip of this s.cience ability to explai n, i.e. to turn the descriptiv.e statements to a variety of (those learned in the definitions) to good use in responding r. These teache questions which will arise for the scientist as inquirer or 0ner, quest1 topic, questions reflect particular presuppositions - about one, oners: questi two relevance, contrast-classes, and so on. Imagine t studen the one and hail, the inquiring scientist seeking to understand t accoun his of basis tial eviden (or rival scientist) asking Aristotle for the fall hail does "Why like ons of hail. The investigator might ask a questi at10n for this more often in all than in winter?" and then seek an explan Aristotle ask might l riv or t studen � phainomenon about hail. The . some ing receiv of at10n expect the with exactly the same question, be well might 0ners quest1 both And hail. of evidence for his account use "Beca r: answe same the from g tandin satisfied by - glean unders . into h.eavy hail is caused when the cool moisture in a cloud is compressed ns. this ice by warmth outside the cloud. " So when Aristotle mentio s of proces explanation in his text, what is he doing - recounting his own inquir y, or citing the evidential base for his discovery? as In other words, I find it plausi ble to regard the Meteorology, and y inquir fic scienti Aristotle has compo sed it, as either a record of real theorebcal research or a listing of the evidence for his proposed conveying es involv science accounts. Furthermore, since teaching uestions w�y-q of sorts of explanations to students who ask a variety e could treatis this etc., es, st-clas with particular topics, relevant contra

(48) It is instructive to compare my view with Br� nschwi? 's claims that it would be _ artificial to describe dialectic, as set forth in the Topics, as either strictly a � eth.od. of discovery or a method of justification. ("Mais, sur le plan historique, ce�te d1�soc1a�1on entre decouverte et justification risque, me semble-t-il, d'introduire certa1nes d1stors1ons dans la lecture des Topiques.} See Brunschwig (1987).

317

very well serve as the text in an ancient class on meteorology. In ascribing to Aristotle this emphasis on the pragmatics of explanation, I do not mean to ignore the role of demonstrations and first-principles within this (or any) science. For Aristotle, the actual discovery of a middle term, i.e. grasping the cause which connects a phenomenon to the first-principles, remains something unanalyzable - a job for induction, nous, quickness of wit. f have not focused on this process here. Instead, I hope to have shown that in this actual scientific treatise Aristotle's procedure departs from the restrictive ideal set forth in the Posterior Analytics, within discussions that could be seen merely as rhetorical flourishes preceding and following upon the crucial business of science (the laying down of demonstrations and definitions). However, these "flourishes" are themselves methodologi­ cally significant. In his preliminary studies of the endoxa, I have argued, Aristotle both focuses his theoretical inquiry by refining why­ questions, and directs his search for empirical data by noting failures and missed predictions of earlier scientists. Secondly, by placing emphasis on signs and proofs in the Meteorology subsequent to his actual definitions of phenomena, Aristotle invokes further explanations of relevant empirical data in arguing for the truth of his proposed scientific definitions.

BIBLIOGRAPHY

Peter, 1983. The Nature of Explanation. (New York and Oxford: Oxford University Press.) BARNES, Jonathan, 1975. Arislolle's Posterior Analyfics. Translated with Notes. (Oxford: Clarendon Press. Clarendon Aristotle Series.) , 1980. "Aristotle and the Methods of Ethics." Revue lnlernationale de Philoso­ phie 34: pp. 490-5 1 1 . , 1981. "Proof and the Syllogism." pp. 2-59 in Berti (1981). BARNES, Jonathan, BRUNSCHWtG, Jacques, BuRNYEAT, Myles, and SCHOFIELD, Malcolm {Editors), 1982. Science and Speculation: Studies in Hellenistic theory and praclice. (Cambridge and Paris: Cambridge University Press and Editions de la Maison des Sciences de !'Homme.) BELNAP, Noel D., Jr. 1963. An Analysis of Questions: Preliminary Report. (Santa Monica: System Development Corporation.) BERTI, Enrico {Editor), 1981. Arislolle on Science: The Posterior Analylics. Proceedings of the Eighth Symposium Aristotelicum 1-Ield in Padua from September 7 to 15, 1978. (Padova: Editrice Antenore.) BoLTON, Robert, 1976. "Essentialism and Semantic Theory in Aristotle." The Philoso­ phical Review 85: pp. 514-544. - , 1987. "The Epistemological Basis of Aristotelian Dialectic." BOYD, Richard N., 1984. "The Current Status of Scientific Realism." Pp. 41-82 in Leplin (1984). AcHINSTEIN,

"I i '

ii ' I!

. l

I . , 'I I I

I

318

C. A. FREELAND

BRODY, Baruch, 1 972. "Towards an Aristotelian Theory of Explanation." Philosophy of Science 39: pp. 20-31 . . . . BRUNSCHWIG, Jacques, 1 987. "Quelques commentaires sur la commun1cat1on de Robert Bolton." i BuRNYEAT, Myles, 1981. "Aristotle on Understanding Knowledge." Pp. 97-139 n Becti (1981). . Barnes, Brunsch wig et , "The origins of non-deductive inference." Pp. 1 93-238 1n al. (1982). CHURCHLAND, Paul M. and 1-IooKER, Clifford A. (Eds.) 1 985. Images of Science. Essay� _ on Realism and Empiricism, with a Reply from Bas C. van Fraassen. (Chicago. University of Chicago Press.) . . of Goodness. ,, The CooPER, John, 1 988. "Review of Martha Nussbaum, The Fragiltly Philosophical Review XCVII: pp. 543-564. . . DEVEREUX, Daniel T ., 1987. " D JaIectIC, 'Th e Method of Endoxa', and Scientific Inquiry." . ,, . Aristotle's FREELAND Cynthia, 1986. "Scientific Method 1n Meteorology. - 1 99 1 "Aristotle on Causal Explanation." Manuscnpt, draft. Forthcom1ng. 1n Linds;y Judson, Ed., Essays on Aristotle's Physics (Oxford: Oxford Un1vers1ty Press, scheduled publication date 1991). . . . , , The Journal of FRIEDMAN, Michael, 1974. "Explanation and Sc1ent1fic Understanding. Philosophy 71 : pp. 5-19. . HEMPEL, Carl G., 1965. Aspecls of Scientific Explanation and other Essays in the Philosophy of Science. (New York: Macmillan, The Free Press.) HiNTIKKA, Jaakko, 1972. "On the Ingredients of an Aristotelian Science." Noiis 6: pp. 55-69. . , 1 976. The Semantics of Questions and the Questions of Semanf1cs. A cta Philosophica Fennica 28, n. 4. . . , 1 979. "The logic of information-seeking dialogues: A model." Erkennlnis 38: pp. 355-368. In D. O'Meara (Ed.), Studies in IRWIN, Terence, 1981. "Aristotle's Methods of Ethics." Aristotle (1981). KAKKURI-KNU UTTILA, Marja-Liisa, 1987. "Aristotelian Games of I_Jialectic." In M. Kusch and H. Schroder (Eds.), Texl-Inlerprelalion-Argumenlatton (Hamburg: &� . with Martin KuscH 1 988. "LSP-Research, Philosophy of Science, a°:d ilie Question-Theoretical Approach - Some Tentative Suggestions." (Manu�cnpt, pp. 1 -52.) KuHN, Thomas, 1970. The Siruclure of Scientific Revolutions. (Chicago: University of Chicago Press.) 1984. Scientific Realism. (Berkeley: Universi�y of California LEPLIN, Jarrett (Editor). Press.) MORAVCSIK, Julius, 1974. "Aristotle on Adequate Explanation." Synthese 28: PP· 3-18. NussBAUM, Martha, 1 982. "Saving Aristotle's Appearances." Pp. 267-293 in Schofield and Nussbaum (1982). . in Greek Tragedy and _ , 1 986. The Fragility of Goodness. Luck and Ethics Philosophy. (Cambridge: Cambridge University Press.) . . OwEN, G. E. L., 1961. "Tilhenai ta phainomena." Pp. 83-103 in S. !'1an�10� , Ed., Aristole et tes problemes de mithode (Louvain: Publications Un1�ers1taires de Louvain). Reprinted as pp. 167-90 in J. Mor�vcsik, Ed., 1967 Artslolle (Garden City, N . Y.: Doubleday) and in Articles on Aristotle vol. 1 (1975), PP· 1 13-26. _ POPPER, Karl, 1962. Conjeclures and Refutations. (New York: Basic Books.) REsCHER, Nicholas, 1982. Empirical Inquiry. (�ondon: The Athlone Press.) SALMON, Wesley C., 1984. Scientific Explanation and the Causal Siruclure of the World. (Princeton: Princeton University Press.)

l

!1·.1 .i :1. 1

i

::I ;''I

T1

I i

1

I

I

Malcolm and

NusSBAUM,

Studies in Ancient Greek Philosophy.

·

I>

319

Martha (Editors). 1982. Language and Logos: (Cambridge: Cambridge University Press.) SEDLEY, David, 1982. "On Signs." Pp. 239-272 in Barnes et al. (1982). S1NTONEN, M. 1984. The Pragmatics of Scientific Explanation. Helsinki: Acla Philosophi­ ca Fennica, vol. 37. TOULMIN, Stephen, 1972. Human Understanding. (Princeton: Princeton University Press.) VAN FRAASSEN, Bas C., 1 977. "The Pragmatics of Explanation." American Philosophi­ cal Quarterly 14: pp. 143-50. , 1980(a). "A Re-examination of Aristotle's Philosophy of Science." Dialogue (Canadian) XIX: pp. 20-45. , 1980(b). The Scientific Image. Oxford: Clarendon Press. SCHOFIELD,

·

i!ii··.1 , !1.

SCIENTIFIC EXPLANATION AND EMPIRICAL DATA

Appendix I: The Salliness of the Sea

I

\

l

Aristotle takes up his examination of the sea in I I , and focuses mainly on why it is salty, a question he associates with issues3 about its origins and composition in general. He explores endoxa including Empedocles' claim that the sea is the sweat of the earth, and criticizes them (this one, by fails to explain, by not giving an account of "sweat"). Finallysaying that it he moves to clear away these accounts and give his own, immediately g to his own assumption of the exhalations (357b24-26). He considers italludin "obvious" that the crucial factor will involve the dry exhalation. Next he notes various "signs" (semeia) that the sea's saltiness is due to some admixture. in particular are mentioned. In living bodies the least digested matter isTwo salty and bitter; and, in combustion, ash remains. These are assimilated as both alike being cases where heat fails to master all its material. He remark s, Just as in combustion there is a residue of earth of this kind, so there is in all natural growth and generation, and all exhalation on dry land is such a residue. Dry land is such a residu.e , and it comprises most of the exhalation, which gets mixed with moist exhalation, turned into rain, anddry subsequ ently rained down and deposited in the sea. In crude syllogistic form we would express Aristotle's reasoning to his conclusion like this: What includes a residue of the hot dry exhalation is salty. The sea includes a residue of lhe hot dry exhalation.

I

]

The sea is salty. Now, j ust as in his treatment of earthquakes, so here too does Aristot le proceed to lend support or credibility to his account by putting to use in explaining associated phenomena (sumbainonla). First he explainsitwhy "rains from the south and the first rains of autumn are brackish." He next account for why the sea is warm and for why it never dries up. In addition, within s this discussion he refers to a variety of homespun experiments all designe d to prove his point, that the sea's saltiness is due to an admixture, one which fails to be present in any vapor formed over the sea. He notes that, first, with wine and other "tasty liquids" (khumoi), any evaporation simply become s water on condensation. Next he describes an experiment where a bottle with a wax stopper thrown into the sea lets in only pure fresh water, "for the earthy substance whose admixture caused the saltness is separated off as though in a filter." Furthermore, he cites it as a "proof" (lekmerion) that water's density becomes greater through admixture that eggs will float in very salty water as bodies float in �he Dead Sea.) Each of these simple experiments is cited Uust in an

�· I I

'

Ji

,I

1.

320

C. A . FREELAND

effort to show that the cause of the sea's saltiness is what he says it is, an admixture of the hot dry exhalation. How would they do this? Presumably by showing that in other similar circumstances phenomena occur which are like those features or sumbainonla of the sea. For example, the fact that eve_n heavy boats will float in the_ sea is taken to be another interestin� pathos _ _ concerning the sea, and one which (I think) Aristotle regards as another sign that the sea's saltiness is due to an admixture; as evidence he reminds us that such a mixture serves to float eggs, on a smaller scale.

1

B10LOGIE, LOGIOUI! ET MllTAPHYSIOUI! chez

Editions du CNRS, Paris, 1990.

ARISTOTLE'S DE CAELO I 12 AND HIS NOTION OF POSSIBILITY

Appendix II: An Anomalous Methodological Remark

' i '.

: · ,,

A very important - though frustratingly brief - methodological remark occurs within Aristotle's discussion of comets, where he seems to present an antirealist position. Here Aristotle suggests that there are varying degrees of testability and, perhaps, also of confirmability within a science like the Meteorology. He says,

MARIO M IGNUCCI

We consider that we have given a sufficiently rational explanation of things inaccessible to observation by our senses if we have produced a theory that is possible: and the following seems, on the evidence available, to be the explanation of the phenomena now under consideration. (344a5-8) Which things exactly are "inaccessible to observation" (aphanon lei aisthe­ sei)? The cautionary remark is made only here; it may be that Ar�stotle

I

intends it to apply primarily to heavenly phenomena, as opposed to those in and above the surface of the earth. At any rate, he does not hesitate to argue that various of his own proposals are correct; above I have examined his employme�t o f various "signs" or "proofs" (semeia or tekmeria) for the correctness of his causal stories. Since he even makes �uch a claim for his proposed account of comets, it is hard to see what the impact is of the qualification made in the" passage quoted just above. I am tempted, though without any further j ustification, to regard it as an interpolation (and one which is Epicurean in tone).49

�i . i

:!I · '. .

i 'I '! :i'

1·. 1.· !i >

(49) See Sedley on Epicurus' views about the equal plausibility of all theories of celestial phenomena which are consistent with the phenomena (Sedley (1982], esp. pp. 26872).

Alusrora

8eminaire CNRS-N.S.F., 1987

The problems raised by Gael. I 1 2 are well known to Aristotle scholars and despite the numerous papers devoted to it there is no substantial agreement on its interpretation. Although it may appear bormg to continue an endless debate, this Aristotelian chapter is so important for its logical and philosophical implications that a reader feels compelled to come back to the old story and make his own judgmenL To underline the relevance of the logical problems involved it is sufficient to remember that our chapter is one of the pillars on which the so calle d ' '. statistical interpretation" of modal operators proposed by _ Jaakko Hmtikka is based.1 As is known, according to Hintikka, Anstotle would have m aintained a view here and elsewhere according to _ _ _ necessary if and only if it which a propos1t10n p is is always true. Of course, one side of this equivalence is quite plausible, i.e. (LA) Lp =l> Vtp (I) I take "p(I)" to mean that p is the case at I. its converse, i.e. (AL) Vtp(I) =l> Lp

The difficult side is

Needless to say, holding (LA) and (AL) is the same as asserting (LS)

Lp = Vlp(I)

(1) Cf. J . Hintikka, Time and Necessity. Studies Oxford: Clarendon Press, 1973, pp. 93-1 13; 151-152. \

in Aristotle's Theory of Modality '

!:

[ I

i \1.I I

1

l i I

322

M. MIGNUCCI

and this allows us to characterize possibility by means of the following equivalence: Mp = Hlp(l) (MS)

'' ''

I would not like to discuss the general question of whether or not Aristotle ever defended (MS). Such a general question is difficult to answer and would involve an examination of many controversial passages which are connected with very complicated historical and philosophical questions. Nobody should expect that this interconnec­ ted nest of questions could be even briefly answered in one paper. Let us take a more modest stance and claim that there is no reason to suppose that anything like (MS) is stated or logically implied by Aristotle in Gael. I 12. That is precisely what I would like to show by inspecting the first part of our chapter. II The general thesis o f the chapter i s well known and represents Aristotle's answer to the cosmological view expressed by Plato in the Timaeus. Needless to say, Plato's position was that the kosmos, the world, cannot be destroyed, although it has been generated. This is at least Aristotle's interpretation of Plato and we know that this interpretation was not shared by many distinguished Platonists even in ancient times.2 Examini ng how reliable Aristotle's interpretation is does not concern us here. His view is that the world is eternal and it cannot be generated or destroyed. For our purposes it is importa nt to consider the first step of his argument in favour of his claim, which is directed to show that whatever is eternal cannot have the possibility of being corrupted. The proof of this thesis is contained in the first part of the chapter and we have to examine it carefully. Let us start by looking at the beginning: (A) (a) Having established these distinctions we can now proceed to the

. !

sequel. (b) If there are some things which can both be and not be, (c) there must be some definite maximum time of their being and of not being; (d) a time, I mean, during which the thing can be and a time during which it can fail to be according to any category, whether the thing is, for example, a man, or white, or three cubits long, or whatever it may be. (e) For if the time is not definite in quantity, but always longer than any given time and shorter than none, (f) one and the same thing will be able to be for an infinite time and not to be for another infinite time. (g) This however is impossible.a

(2) A discussion of this problem can be found in II. Cherniss, Aristotle's Criticism of Plaio and the Academy, I, New York: Russell & Russell, 19622, p. 421 ff. (3) Gael. I 12, 281a28-b2 (revised Oxford translation slightly modified). ! i

I

I

'

I i ' ' '

NOTiON OF POSSIBILITY AND

DE CAELO I

12

323

Before trying to understand Aristotle's move, some preliminary remarks must be made. First of all, it must be pointed out that the temporal restriction which is hinted at in (c) and (d) has to be taken as a restriction on the possibility of being or not being or being such and not being such, and not as a limitation on a subject's having this possibility. If a has the possibility of eating and not eating, what Aristotle claims is not that a has only some of the time this possibility, but that this possibility cannot be a possibility of always eating and always not eating . This point. has been convincingly made by Lindsay Judson in his valuable paper on this subject and I do not need to repeat his argument. 4 We have to establish lhe meaning of "E\lia. 8u\la."t'& xctt eI\la.i xctl µ�" in {b) . Several problems are interwoven. First of all, does "3uv��oc" refer to possibilities or capacities? This question can reasonably be proposed only if a distinction between "capacity" and "possibility" is at work . Suppose I.ha!. any capacil.y implies a possibility and that the converse does not hold. If this is true, I do not see any reason to restrict Arist.ot.le's reasoning to capacities. In what follows he discusses the case of something which may not be. It seems slightly odd to interpret this as a statement that something has the capacity not to be . And it is even stranger to speak of the capacity of being three cubits long, while it is much more natural to say that something is possibly three cubits long. But one might reply that there are contexts in which it may not be strange to say that something is capable of being three cubits long. However, what is relevant is that Aristotle does not speak of the Mv�µ,, to be as opposed to the 3ov�µ,, not to be. He makes his point about what is 3uv��6v to be and not to be, and it appears natural to me to interpret it as a statement about what can be and not be. We will return to this question later on. Besides, it does not seem to me t.hat Aristotle normally makes a clear cut distinction between the two notions. The attitude he takes in Melaph. El 3 is in my view paradigmatic. He starts by considering possibilities which are capaci­ ties, such as the capacity of building.' After a while he slips more or less tacitly into a notion of possibility which we would be inclined to represent by means of the familiar modal operator' and he ends his

(4) Cf. L. Judson, "Eternity and Necessity in De Caelo I 12. A discussion of Sarah Waterlow, Passage and Possibility: A Study ofAristotle's Modal Concepts", Oxford Studies in Ancient Philosophy, I (1983), pp. 226-227. The best evidence for this claim is offered by the discussion of capacities at Gael. I 1 1 , 281a7 ff. A similar point is made also by C. Campus & M. Mariani, "II principio di pienezza in 'Met'. 8 4 e 'De Coelo' A 12", Teo,ia 6 (1986), p. 185. (5) Cf. Melaph. 0 3, I046b29-I 047a l0. (6) Mclaph. ,e 3, I047a l0-14.

l

324

NOTiON OF POSSIBILITY AND

M. MIGNUCCI

is discussion of the Megaric position by a definition of possibility which his h althoug ore, Theref his usual general definition of possibi lity.' to examples of 3uvocµe"; at Gael. I 1 1 , 28la7ff. seem to apply better is point his that this from follow capacities than possibilities, it does not l genera include not does and restricted to mere capacities we passage the after iately immed possibilities. On the other hand, some have just quoted Aristotle offers a kind of digression in which that us s_ remind He stated. are theses and usual modal distinctions be must w<; \moOfoe <� falsity and truth lity, impossibility, possibi his if Even .8 notions e absolut onding corresp the distinguished from a illustration of the distinction is far from clear,9 there is no doubt that the of modal doctrine is hinted at which is familia r to any reader what is Corpus. For instance, Aristotle distinguishes repeatedly , with Besides &:v\lyxa.tov E:� U7to6�cre:co<; from what is OC7tA&<; &:vixyxoctov.10 nes underli he way e respect to what is false or impossible in an absolut am I that say to : that we have to separate clearly the two notions the on ible; imposs not but standing when I am sitting is simply false nsura­ other hand, the statement that the diagon al of a square is comme ,a Finally ible.11 imposs simply also ble with a side is not only false but lent equiva an in ere elsewh found be can modal thesis is mentioned which ible, B form," namely the thesis that if B follows from A, and A is imposs which of 3uvocToc the that le plausib it makes this All is impossible too." can which affairs, of states e possibl just are g Aristotle is speakin the by adequately be expressed by propositions modall y qualified operator of possibility. Aristotle is then speaking of the possibilily of being and not a being. This possibility of being and not being is in my view ally tempor be must it possibility of existing and of not existing. Since al limited, the existence which Aristotle has in mind is a tempor ,can We ce. existen logical existence, which is different from purely if a is represent the latter by stating a la Quine that a exists if and only only and if is that se, discour of e univers something which belong s to our if :>Ix(x = a) ( I) g in holds. On the other hand, temporal existence refers to a's existin ·

(7) Melaph. 0 3, 1047a24 ff. (8) Gael. I !2, 281b2-8. is (9) For instance that the diagonal of a square is commensurable with the side an as taken is it while b6, 281 at lnco0S:o-ewi; E� lity quoted as an example of impossibi instance of &:nA&i; &:vixyxct.tov few lines after (281bl2-13). (IO} E.g. Ph II 9, 1 99b34ff and H. Bonitz, Index Arislolelicus, Graz: Akademische Druk- u. Verlagsanstalt, 19552, 42b1 0 ff. (11) Gael. I 12, 28lb6-14. (12) Cf. Meiaph. 0 4, 1047b14-26; AP,. I 15, 34a5-33. (13) Gael. I 12, 28lb15.

DE CAELO

I 12

325

an interval of time. We may express this by means of a dyadic predicate "E!(x, y)", so that we can state e.g. (2)

E!(a, j)

to say that a exists at j or in the interval j. It is important to notice that Aristotle in (d) extends his point about the possibility of existing . and of not existmg to any kind of phasal predicate. We have to put a temporal limit not only to the possibility of existing and not existing but to any possibility of being such and not such a I f this is true, we must co be prepared to admit that Aristotle's claim covers any possible state of affairs, and not just the possibility of existing and not existing. For any state of affairs of which the possibility of both its taking place and not taking place is asserted we have to introduce a temporal restriction. Let us further delay the analysis of Aristotle's argument by trying to answer a further question. If "3uvocToc" implies a reference to a modal operator, how should we conceive of it? There are prima facie two ways of mterpretmg the notion of possibility involved. One is by taking it as expressing what is normally called "strict" or "open possibility". As is known, the standard notion of possibility is opposed to impossibility, . but it does not rule out necessity. Let us call this possibility "broad possibility" and represent it by "M". Aristotle sometimes takes possibility in a stricter sense by opposing it not only to impossibility, but also to necessity. In other words, p is strictly possible, i.e. Ep, if and only if p is M-possible and ""1p is M-possible.15 We have formally (ED)

Ep <=* Mp A M-,p

�arah Waterlow's claim that Aristotle's thesis is limited here to predicates . �(14) wh1c !�ll 1n one of the categories has no support in the text (cf. S. Waterlow, Passage and Poss1b1 Lily. � Stud!! of Arislo�le'� Modal Concepts, Oxford: Clarendon Press, 1982, pp. 58, � l). ' '�ix61 o7tota:\louv x ni�opto::v does ?ot mean "according to any category", but � according to any predication" or "predicate". Another criticism of Waterlow's claim can be found in Judson, pp. 222-224. (15) This way of conceiving possibility is reflected for instance in the definition of �v�ex6µevov at APr. I 13, 32a 18-20. The distinction between E - and M - possibility is w1? ely � cknowled�ed in the modern literature on Aristotle. Cf. e.g. A. Becker, f)ie ar1stolel1sche Theorte der M6glichkeilschliisse, Berlin: Junker und Diinnhaupt Verlag, 1933, pp. 7-1 1 ; I. M. Bochenski, La logique de Theophrasle, Fribourg en Suisse: Librairie de l'Universite, 1947, pp. 68-69; D. Ross, Aristotle's Prior and Posterior Analylics A Revised Text wilh !nlroduclion and co�menlary, Oxford: Clarendon Press, 1949, p. 327 ; . . , _ _ J Lukasiewicz, Artslolle s Syllog1sltc from lhe Slandpo1nl of Modern Formal Logic, Oxford: Clarendon Press, 19572, p. 191 ; S. MacCall, Aristotle's Modal Syllogisms, Amsterdam: North Holland, 1963, pp. 66-70; 1-Iintikka, pp. 27-31; G. Granger, La theorie arislotelicienne de �a science, Paris: Aubier, 1 976, pp. 179-182; U . Wolf, MOglichkeil und Notwendigkeit bei Artsloleles und heule, MUnchen: Fink, 1979, p. 102; G. Seel, Die Arislolelische Modallheorie' Berlin: de Gruyter,, 1 982, pp. 158-160; Waterlow, pp. 16-17. ·

:!

326

NOTION OF POSSIBILITY AND

M . MIGNUCCI

It is clear that it cannot be this kind of possibility which is in it question in our passage. There is no contradiction in supposing that and time te mdefim an for place takes is possible that a state of affairs that it is possible for its opposite to take place mdefimtely. . The possibility which is at issue here is something d1ffere� t and it concerns the possibility of a state of affairs and its_ opposite s t_akmg of place. The point is made by Aristotle very clearly m the followmg says: He (A). the chapter, when he comments on the main thesis of

(B) A man has, it is true, the possibility at once of sitting and sta.nding, it does because when he possesses the one he also possesses the other; but . f�re�; 1f � at but stand, and sit m i t me s the � � not follow that he can at _ ty, . times But if a thing has for infinite time more than 1one poss1b1h another time is impossible and the times must coincide. 7

Aristotle's claim is simple. Suppose that we have two states of affairs represented by the following propositions: (i)

Socrates is sitting

and (ii)

Socrates is standing

· It is surely possible for Socrates to sit and to stand, but not at the same time. With respct to (i) and (ii) it is easy to express Aristotl e's point by imagining that (i) and (ii) are implicity tensed and by statmg that it cannot be possible that they are both true 1f they refer to the same time. Therefore, let "p(j)". be "Socrates is sitting at j" and "q(k)" "Socrates is standing at k". A necessary condition for the truth of M(p(j} A q(k)) (3) is that j is different from k, since --i M(p(j) A q(j)} (4) surely holds for any j. 18 , We are now in a position to understand Aristotle s m� �n Lh ��1s in text (B). If we are dealing with propositions such as (1) or (11) (or perhaps better with the states of affairs denoted b_y such propos1t10ns) we cannot take p(l) as referring to an indefimte time or, which is the same, to any time. In other words, suppose that we have two propositions p(l) and q(l) which are inconsistent in the sense that they •

.

•

.

(16) As Judson, p. 229 ·has nic� ly remarked, Aristotle's way o! speaking i� infelicit�us here, because it gives the impression that we have to do not :' 1th a c� pac1ty of doing something for ever, but with the eternal possession of a capa c1ty of do1�g. _ shghtly _ (17) Gael. T 12, 281a15-19 (revised Oxford trunslation modified). (18) That -i(j = k) is just a necessary condition for the t: uth of {3) can be seen by reflecting that (3) is false if an impossible proposition is substituted for P or q.

DE CAELO

I 12

327

satisfy (4). To take them without any temporal limitation means to state Vlp(l) and Vlq(l). It is obvious that --i M( Vtp(l) A Vlq(l)) (T) holds. (Ti )

The same is true is we take as "q" the negation of p, since --iM( Vlp(l) A Vtnp(l))

is surely a perfectly safe law of standard modal logic, since the set of instants or intervals of time l is supposed not to be empty by hypothesis. If this interpretation is correct, there is no reason to endorse Lindsay Judson's daring view, according to which the 3uvocµ«' which Aristotle is considering here are the capacities of the eternal elements which constitute the cosmos by their ordered arrangement. His reasoning is as follows: If a t�ing has the capacity to exist for a limited time only, then how can its capacity for non-existence be for a limited time also? What does the thing do when both its capacities have been fully exercised?19

From this remark Judson concludes that in Aristotle's passage eternal but non perishable individuals are at issue, and they must be the cosmic elements.20 Against this argument an objection can be raised. Aristotle does not claim that the capacity of being is opposed to the capacity of not being. That of which is speaking is the possibility of being and not being. It is not the capacity of Socrates' sitting against his capacity of standing which is in question here. It is his possibility of both standing and sitting at the same time which is denied. Therefore, our text does not rule out that there is a capacity or possibility of indefinitely not being. What it does dismiss is that there is at the same time a possibility of being and not being for an indefinite time. To put it in terms of capacities, our text just excludes that it is possible for a thing to exercise for an indefinite time one capacity and its opposite. And of course this applies not only to eternal, but also to contingent things.21 III So far so good. The difficult part o f Aristotle's reasoning i s the application of (T1) which is made by him to show that what is eternal is not destructible. The essential part of his argument is contained in the following passage:

(19) Cf. Judson, p. 226. (20) Cf. Judson, pp. 226-227. (21) For a di fferent criticism of Judson's position see Campus-Mariani, pp. 185-186. ,

.J

d , iJ \: 'I ' · : .1· :

i ii I

(C) (a) Thus if anything which exists for infinite time is destructible, it will have the possibility of not being. (b) Now if it exists for infinite tir:ie let its possibility be actualized; (c) and it will be in actuality at once existent and non-existent. (d) Thus a false conclusion would follow because a false assumption was made; (e) but if what was assumed had not been impossible its consequence would not have been impossible.22

I !

'!

The argument starts with a conditional whose ground is offered by the discussion of the uses of "indestructible" and its opposite " destructible" put forward in the previous chapter. The proper definition o f "destructible" is the following: are (D) (a) The uses of the words "destructible" and "indestructible" and similar. (b) " Destructible" is applied to that Which formerly was afterwards either is not or might not be whether a destruction and change tible" intervene or not at some time. (c} In some cases we call "destruc we that which may not be because of a process of destruction. (d) Besides, d" destroye "easilythe ible, destruct easily is which that call "destructible" so to speak.23

It is clear that the sense of " destructible" which is in question in ed by text (C) is what is described by clause (b). This is confirm passage the after lines few a (b) n definitio recalls who Aristotle himself e­ quoted in (C) at Gael. I 12, 28l b27-28. I take the disjunctive charact tible rization of " destructible" in (b) as a proper disjunction : a is destruc I' if and only if a exists at I and either (i) there is a time I' such that I < in I > I' time a is and a does not exist at I' or (ii) it is possible that there (ii) which a does not exist at I'. In this view the fulfillment of conditi on is sufficient to call something destructible. Let me insist on this point. In (b) Aristotle explicitly states that as actual destruction is not necessary to qualify something to e referenc l essentia an that nes destructible. Besides , he underli impossibility is contain ed in the proper meanin g of "indestructible "; the reason why a thing is properly called "indestructible " is that it cannot it undergo destruction.24 Finally , if something were destructible only if be would (C) text of nt argume whole the time, some at is destroyed trivial. If a is eternal, i.e. always exists, a is indestructible,, because it does not undergo destruction at some time ex hypolhesi. What makes Aristotle's argument non-trivial (although wrong) is precisely the idea that a thing can be truly said to be destructible even if it is never destroyed. The statistical interpretation of the modal operators seems to be far removed from Aristotle's present way of putting things. However, let us return to text (C) and to Aristotle's proof that what is eternal is indestructible. To say that a is eternal amounts to stating Gael. I 12, (23) Gael. I 1 1 , (24) Cf. Cael. I

(22)

i) 1

I' : '

I'I iI.'.1' Ill

;

M. MIGNUCCI

328

281 b20-25 (revised Oxford translation slightly modified). 280b20-25 (revised Oxford translation slightly modified). 1 1 , 28la1-4. I do not see ho\v Hintikka, p. 151 note 9 and p. 102 can claim that a purely temporal characterization of "indestructible" is given by Aristotle.

NOTiON OF POSSIBILITY AND

DE CAELO

I 12

329

tha a always exist� and we can represent it by VIE!{a, I). To express _ slightly more complicated because we need something _ desoruct1b1lity IS like (DD)

D(a) iff 3't(E!(a, I) A M 3'1'(1 < I' A -iE!(a, I'))"

where, of course, "D(x)" stands for "x is destructible". main thesis of Aristotle in (C) can . be expressed by (MT*)

Therefore, the

VIE!(a, I) =:> -i( 3't(E!(a, l) A M 3'1'(1 < l'A -iE!(a, t'))

As (a) clearly suggests, (MT*) can be simplified to (MT)

VIE!(a, I) =:> -,M 3'1 -iE!(a, I)

since (MT) implies (MT*). The proof of (MT) is by reduclio ad absurdum. This means that we to suppose that the negation of (MT) is true, i.e. (5)

VIE!(a, l) A M 3'1-iE!(a, I)

In (b) a further assumption is made, i.e. that the capacity of a is actualized. I take 1t to mean that the possibility for non-existence of a IS supposed to be realized, i.e. that a certain time a does not exist. We will evaluate this move later on. Suppose then that 3'1 -iE!(a, t) is he case. With respect to it Aristotle rightly points out that it cannot be true together with \flE!(a, I). In fact we have -iM( \ftE!(a, l) A 3'1-iE!(a, I)) (6) as a special case of (T,). Therefore - this is Aristotle's conclusion since we have reached something absurd from supposing that a . sometimes does not ex1St, 3'1-iE!(a, I) cannot be possible. Thus, if a is eternal, there cannot be a time in which a does not exist, and this means that a is indestructible. This conclusion must carefully be examined, smce it 1s far from obvious. (6) is not the negation of (5) and Aristotle develops an argument in order to go from (6), which is indisputable, to the negatwn of (5). This argument is based on the idea that it is not sufficient to say that hE!(a, I) is false, but it must be added that it is impossible, since it yields a contradiction. A possible way to explain Aristotle's move is by supposing that

(25) Susanne Bobzien has suggested to me that it would be better to put the modal operator aft.er �he �emporal quantifier, since the possibility of time is not in question here. But if time is supposed to be one and the same in the different counterfactual sit� ations with .respect to which we evaluate modal propositions, we can think of it as a set of 1nst�nts ?r . intervals which is one and the same in different possible worlds. If so, something s1m1lar to the well known "Barcan formula" can be assumed, which allows us to reverse the order of the existential quantifier over time intervals or instants and the modal operator of possibility. I prefix the modal operator to the existential quantifier for technical reasons.

NOTiON OF POSSIBILITY AND

M . MIGNU CCI

330

s that his when he says that :>IhE!(a, t) is possib le, he implie true. In be to actualization is consistent with whatever is supposed and B is true other words, Aristotle's view seems to be that if A is as a state ore theref possib le, then B can be true together with A. Let us general law A A MB =!> M(A A B) (R) ,,

. r ,,

(8)

''

I I i, :, '

i' ::

' :·

a To see this it is sufficient to substitute "1A" f?r "B" in (R) . .There is surely true be cannot it But false. being of ty possibili its with together true case in which A is that A is true together with its negation. (26)

I

331

M :>It>E!(a, I) ='> :>ihE!(a, I)

would be perfectly all right and there would be no reason for Aristotle to say that :>ihE!(a, I) is false if he had shared the statistical interpretation of modalities.29 Besides, what we have already hinted at must be underlined. The possibility which plays a role in the definition of "destructible" can hardly be reduced to its temporal interpretation. Aristotle clearly says that a thing is destructible independently of the fact that it undergoes a destruction. But in a temporal perspective if a is never destroyed it cannot be said to preserve its possibility of being destroyed. Finally, the statistical interpretation is strongly reductionist as far as modalities are concerned. Modal operators are simply replaced by quantifers over instants or intervals of time. In this frame thc thesis according to which what is eternal is indestructible becomes trivial and one can hardly understand why Aristotle presents so elaborate an argument in order to prove simply that if a always exists there is no time in which it does not exist. However, a defender of the temporal reading of modalities might insist on his view. Nobody can deny that Aristotle asserts (MT). (MT) can easily be generalized to

IV not Insisting on underlining the fallacy of Aristotle's argument does which ct10n r matter. It is more important to stress that in the reconst � statistical we have made there is no need to resort to Hmtik ka s sion as conclu this to interpretation of modalities. One might object the way this m ent argum follows: all right, by interpreting Aristotle's have we pnce the But role. statistical interpretation does not play any te a to pay for such a manoeuvre is that we aPe compelled to attribu _ ical statist the if d avoide be would fallacy to Aristotle. This charge _ we le prmcip charity the by ore, Theref interpretation were admitted. . etation interpr . must prefer the statistical . I believe that the charity principle must be adopted m readmg of the ancient texts, but it cannot have as a consequence a distortion ed propos historical facts. In my view the interpretation which I have the by out is deman ded and the statistical interpretation ruled le in (d) texts. Let us return to text (C) and point out that Aristot that says he cisely, pr qualifies as false the assumption made. More � . a yields it smce ible, this assumption is not only false, but also imposs

I 12

contradiction. It seems inevitable to conclude that the false assump­ tion of which Aristotle is speaking is that a sometimes does not exist, i.e. :>lhE!(a, t). In what sense is this proposition false? A natural answer is that :>If--,E!(a, I) is qualified as false because it does not follow from M :>/f--,E!(a, t). This way of speaking is quite common in Aristotle and several passages can be quoted where the same point is made - for instance at APr. I 15, 34a37-38, where a similar argument is proposed in order to justify Barbara XEM,27 as we will see in detail later on. 28 Suppose now that the statistical interpretation holds. Possibility must be defined according to (MS) and this means that it must be eliminated in favour of an existential quantification on the instants or intervals of time. From this point of view an implication such as

By applying (R) to our case we get !(a, I)) VIE!(a, t) A M :>ihE!(a, I)) =!> M( VIE!(a, l) A :>ihE (7) n imme diately Since the consequent of (7) is the negation of (6) we c� the negat10n 1.e. dent, antece its derive by contraposition the negation of uently conseq and (MT) assert to of (5). Therefore, we are entitled �T �. . , ds on Interpreted in this way, the whole of Aristotle s proof depen fore, There see.26 to easy is as d, invali is (R). Unfortunately, (R) _ does not prove (MT). And we it and ect incorr Aristotle's argument is tle's wrong have to expect this, since (MT) is obviously false. Aristo it must be idea depends on supposing that if somethmg is possible, . is the which possib le in such a way that it is consistent with everythmg case.

' '

DE CAELO

(MT a)

II

!

I i

Vtp(t) ='> L Vtp(t)

(27) "Barbara XEM" means "Barbara with a categorical major and strictly possible minor premiss and broadly possible conclusion". (28) It must be pointed out that the interpretation of "false" I am illustrating here was first proposed by D. Ross, pp. 338-339. (29) I do not understand how 1-Iintikka, p. 190 can claim that APr. I 15, 34<137-38 just quoted is a piece of evidence for his temporal interpretation of modalities. Ry assuming that "it is possible for B to apply to all C" Aristotle is not committed to the view that "at some moment of time it is therefore true to say 'B applies to all C' " . If this was his positions, why , should he have qualified this latter proposition as false?

l 332

M. MIGNUCCI

since the same reasoning which gives (MT) can be applied to any proposition which is the case at any time. And, of course, (MT G) is a special case of (LA) which is the difficult side of the statistical characterization of necessity. Therefore, Aristotle in the De Caelo passage endorses a temporal view of modalities. Well, it cannot be denied that Aristotle by asserting (MT) implicitly states (MTG) and therefore some form of a temporal interpretation of necessity. The situation is however quite different from the. way in which Hintikka portrays it. He claims that Aristotle's asserting (MT) in the De Caelo is a consequence of his general view about modalities. On Hintikka's account, since Aristotle identifies " necessa­ ry" with "always true", he draws the consequence that whatever always exists it is so necessarily. In my interpretation Aristotle's conclusion is not a consequence of a general view about modalities, but the result of a mistake about the laws of distribution of possibility. This means that (MT) is not a thesis of a consistent theory of modalities, but a non­ sequilur, i.e. a proposition which he would not have asserted if he had been consistent in his views about modalities. Therefore, (MT) does not hint at an alternative conception of the modal notions, but simply shows how is easy to get mistaken when modalities are in question. v

11.·. ·

I hope that the interpretation l have proposed is plausible. If we . accept it however, we have to charge Aristotle with a modal fallacy. And of course Aristotle's partisans might find this view quite repellent. Not just to soften their reaction, but to insist on the obvious observation that we must read Aristotle, like any other philosopher, without prejudices, it must be pointed out that there are other passages in which Aristotle is guilty of the same logical mistake we have attributed to him in De Caelo I 12. Let us consider APr. I 15, 34bl9-27. In this chapter Aristotle considers first figure modal syllogisms which have one categorical and one possible premiss. The passage which interests us concerns Celarenl with categorical major and possible minor. Aristotle's claim is that its conclusion is a M-possible proposition. Therefore, he believes that Celarenl XEM holds good.30 His argument is the following: (E) (a) Again let the proposition AB be universal and negative, and assume that A belongs to no B, but B possibly belongs to every C. These being laid down, it is necessary that A possibly belongs to no C. (b) Suppose

(30) As usual, "X" indicates that the major premiss of Celarenl is a categorical proposition, "E" that its minor premiss is stricly possible and "M" that the conclusion is possible in a broad sense.

NOTION

OF POSSIBILITY

AND

DE CAELO

I

12

333

that it cannot belong, (c) and that B belongs to C, as above.a• (d) It is ne?essary then that A belongs to some B; for we have a syllogism in the third figure. (e) But this is impossible. Thus it will be possible for A to belong to no C; for the �onsequence of a false assumption is impossible (APr. I 15, 34bl9-27; revised Oxford translation slightly modified).

Sentence (a) reports the thesis which Aristotle wants to prove, i.e. Celarenl XEM, and we can express it by (CX) AeBA EBaC =;. MAeC As (b) clearly shows, Aristotle 's argument is by reduclio. Suppose then that (CX) does not hold. In at least one case it happens that the premisses of (CX) are true together with the negation of its conclusion, i.e. (9)

AeB A EBaC A LAiC

By substituting BaC for EBaC Aristotle construes the following third figure syllogism: (DL) LAiC A Bae ""'> A iB which is acknowledged by him as valid.32

From it

(10) --iM(AeB A BaC A LAiC) follows immediately, if we admit a sort of necessitation rule." Of course, (10) is not the negation of (9) and what we need to prove (CX) is . not (10) but the negation of (9). Clause (e) is supposed to provide a way to pass from (10) to the denial of (9). If we resort once more to (R) we have an easy bridge to (9). Since Ep ==;. Mp

(1 1)

holds in general, from the hypothesis that BaC is strictly possible, i.e. . EBaC, we derive that it is simply possible, i.e. MBaC. Therefore if (9) ' holds, we may assert also (12)

AeB A MBaC A LAiC

By applying (R) we get (13)

AeBA MBaC A LAiC ==;. M(AeB A BaC A LA iC)

Of course, the consequent of (13) is the negation of (10). can state (14) --i(AeB A MBaC A LA ic)

Therefore, we

(31) Aristotle is here referri�g to the proof of Barbara XEM which is expounded at 34a34-b6 and uses the same device as the one for Celarenl XEM. (32) Cf. APr. I I I , 3I b3!-33. (33) We ea � alternatively take (DL) as a necessary implication, as &.vo:yx1J at 34b23 allows us to admit. In this case (9) should be prefixed by a possibility operator. The argument however ,does not change.

. i, '.

; ;, ' · 1 :.

"

'

334

BIOLOGIE, LOGIQUE Er Ml!TAPHYSIOUE

M. MIGNUCCI

chei: ARisroTE

SCminaire CNRS-N.S.F.,

From (14) we derive immediately --i(AeB A EBaC A LAiC) (15)

' ' i

. and this comp letes the proof of (CX). . . nstration and Once again (R) plays a role in an Anstotehan demo 1ple. And in fact once again we have to repeat that it is an mvah d prmc _ with respect t.o place takes ion situat same The (CX) does not hold." r argument is the proof of Barbara XEM, where a simila r, smce �t 1s paral lel used. Reporting the proof in detail does not matte amme d. 3 to the proof of Celarenl XEM we have already ex. matter, I am If I may venture a conclusion in this complicated not offer any do 15 I inclin ed to say that De Caelo I 1 2 as well as APr. of Anstotle, part the on s new hint at a different analy sis of moda l notion trapped m be can s wa he as but simply show how even as great a logician . . all this. m g exc1tm g nothm the snares of moda lities. And there is

nt for (CX) is corre�t, b.ut (34) Ross, Pp . 339 and 341 thinks that Aristotle's argume . · · d eny1ng 1ts. in k' s view denn1c Tr and 's � . he is surely wrong and we must endorse Becker II. Tredennick, London. by tcs, Analyi Prior le, Aristot and 52-54 pp. , Becker validity. Cf. lleinem ann, 1938, p. 270 note d. (35) Cf. APr. I 15, 34a34-b2.

1987

Editions du CNRS, Paris, 1990.

BIPARTITE SCIENCE IN ARISTOTLE'S BIOLOGY WOLFGANG KULLMANN

The conception of a bipartite science (E:n:l TOC<; &p:x,6:<; - &.n:O -rW" &:pz&v,
(1) W. Kullmann, Wissenschafl und Mefhode. Inferprelalionen zur arislofelischen Theorie der Naturwissenschafl, Berlin-New York (de Gruyter) 1974, eh. I I I : Phanomeno­ logie und Atiologie, p. 154-268. (2) W. Kull�ann, Wissenschafl und Melhode (see above n. I) I65ff.

I

!

! 1_·

336

'

�: , I '

r.

BIPARTITE SCIENCE IN ARISTOTLE'S BIOLOGY

KULLMANN

iI; !i·

!I I 1, : : I

i ;. I

],. ,j ' l1-:!·_ : i I,; ,

'

I

337

te premisses, especially from definitions or definitional propositions ("the planets are near"). Aristotle does obviously not mean to say that the 8T,, i.e. that part of a science which is concerned with facts (provided that it is a science which contains facts as well as reasons) only consists in syBogisms. Among the cases in which the "that" and the "why" are divided among two sciences, an instructive example can be found which proves this point: According to 78 b 35ff. the "phenomena" are separated from mathematical astronomy forming a special discipline _ subordmate to it. The passage is difficult. It is obvious, however tha t th e investigation of astronomical phenomena represents a disciplin� _ purely descriptive, a discipline, which collects facts concerning _ is which _ celestial bodies and is probably identical with "nautical" astronomy. Compare Anal. pr. I 30.46 a 17ff. where the same example occurs, although the term 8T, is lacking: "Therefore it is the task of experience to give the principles which belong to each subject. I mean for example that it is the task of astronomical experience to supply the principles of astronomical science (for it was not until the phenomena had been sufficiently apprehended that the astronomical demonstrations were discovered); and the . same apphes to any other art or science. So if the attributes of each thing are apprehended, it will be our task to exhibit readily the demonstrat10ns. For if none of the true attributes of the things has been omitted in our inquiry, we shall be able to find out and to demonstrate everything which has a proof, and to make that clear whose nature does not admit of proof." (tr. Tredennick, Jenkinson, with changes) 0L0 T0Cc; µE:v &.pxOCi; T0Cc; ne:p1 ExaO'"t'O\! Eµne:tplai; €0'T1 napaOoU\lcxL, AEyw O' oio'J "t'�\I &.a-rpoAoyLx�v µE:v £µ7te:tpla.v -riji; &.cr-rpoAoyixl)i; €7tLcr-r�µ"f)c; (A1)q:>6EvTW\! yOCp lxa.v&i; T&v q:>atvoµEvwv oUTwc; e:UpE61)<J<X\! a.1 &.cr-rpoAoytxat &.7to0e:ll;e:Li;), Oµolwi; OE xa.t 7te:pt &AA"f)v 07toLa.voUv Exe:i -rEX""f)" -re: xat €7tLcr-r�µ1)v · &cr'r' S:OCv A1Jcp67j TOC U7t&.pxovTa. 7te:p1 " ' ' '\:' 't:' .,. e:xa.crTov, 1)µe:Te:pov 1JO1) -ra.i; a.nooe:L�e:ti; e:Totµwi; e:µcpavt-,e:Lv. e:L yap µ1)0£..,. xaTiX -rfiv 1crTop£a...,. 7ta.pa.Ae:tcp6e:l1) -rWv &.A1J6&i; U7ta.px6v-rwv -rot."i; 7tp&.yµa.crtv, El;oµe:v 7te:p1 &navToc; o?J µE:v Ecr-rtv &.7t60e:Ll;ti;, "t'<X6"t'1)\! e:Upe:t."v xa.1 &.7to0e:tx\16\lat, o?J OE µ� 7t€cpuxe:..,. &.7t60e:tl;ti;, -roU-ro 7tOte:t."v cpa.\le:p6v.

an �v3o�ov, i.e. According to his view homology should be based upon or to the wise men most to men, all to so something which seems to be g from �v3o�"' ' startin is which tics, dialec of men. It belongs to the tasks . means, it is no to guide to the &.pxocl of the sciences ( I 2.101 a 36ff.). This '"o<;. On the longer Socrates who is leading someone to the &.px� &.vu7t66 at new arrive and other side the different epistemai start from the &.pxocl g leadin the ally conclusions. Now Aristotle realizes that lmocywyfi, origin with do to. hing somet of a pupil to the universal and the true,3 has syllogism, xaTOC perception. " Induction" is, he says1 in comparison with to perception ing accord -rljv ai0'61)0'L\! yvwptµWTe:pov , more knowable position. presup e decisiv its (I 12. 105 a 17), though perception is not theory of his to comes he when Aristotle silently changes his view former his e revok itly explic not does apode ixis in the Anal. post.• He clear es becom it but tics; dialec by opinion concerning the role played the to sight) first at seem might it (as e that he no longer confines scienc from i.e. , sitions propo first from ds procee apode ictic syllogism which deal with the principles of proof. Chapters I 16-18, for examp le, apodeixis is question in which cases the result of syllogistic demonstration faulty. Aristotle here expressly acknowledges that that epagoge depends on epagoge, i.e. on the leadin g to the universal, and e more becom itself presupposes perception. The terminology has d to relate longer no abstract; the concepts used in this connection are There . Topics the of persons as they were in Plato and in the Aristotle g some pupil is no question of agreement between interlocutors, of guidin Dialectics dy. anybo to right to the univer sal, nor of what seems to be principles first the to g guidin of on has obviously lost its auxilia ry functi means of by them to cted condu be of the sciences. One can only are the which 36�"''• of means by not perception (I 18.81 b 5 o7tocx6ijvoc,), starting-points of dialectics. 5 Here In this respect chapter I 13 is of special importance. edge of the Aristotle investigates the difference between the knowl He states that "that" and the knowledge of the "why" (cf. also I 27). same science the both types of knowledge are sometimes found within ates the elucid tle and sometimes belong to different sciences. Aristo forms of nt differe the difference between the two types by pointing to i.e. the , "that" the of syllogism which they entail . The syllogism ed (in observ ically empir be syllogism of fact, proceeds from what can the of ism syllog the oned); menti 78 a 34f. epagoge and perception are iaimmed from i.e. ples, princi ictic "why" , however, proceeds from apode

1 11 '

W.

"

•

I

\

•

'

,

'

I

In this passage Aristotle emphasizes the universality of the _ method and goes into the preconditions for the application syllogistic of this method. The &.pxocl, he says, are supplied by <µ"ecpfo<. As an example he mentions astronomical oµ7tecploc, which supplies the &.pxocl for the astronomical science. When the phenomena of the stars had been sufficiently collected, he points out in 46 a 20ff., the astronomical proofs were discovered; the same applies to any other lechne or science. Aristotle goes on by saying that as soon as the (mapxonoc, i.e.

York 1942, 75f. {= Der {3) Cf. E. Kapp, Greek Foundations of Traditional Logic, New 89f.). 1965, en Gotting n, Grieche den Ursprung der Logik bei {4) W. Kullman n (see above n . 1) l69ff. {5) W. Kullman n {see above n. I) 206ff.

I

I

/.

1

338

W.

KULLMANN

the properties of any one subject, are fully discovered, it is our task to produce the proofs. I f in the course of our investigation {laTopfot) no true properties of the things have been left out, we are able to find a proof of everything which is provable, and to show what is improvable: With regard to this passage some points should be stressed: Science is based on experience, not on opinions (36�oct). Experience supplies the facts, \,,t
1 ·: ' .1,, i·!,: i: I I

l, i: , I

Hi'.!. ' .

I

11 1

,

i

.•

'

(6) He was no professional astronomer. De Caelo I-II only deal with celestial physics. (7) W. Kullmann, "Die Funktion der mathematischen Beispie\e in Aristoteles' Analylica posleriora", in: Aristotle on Science: The "Posterior Analylics". Proceedings of lhe Eighth Symposium Arislotelicum held in Padua from September 2 to 15, 1978, ed. by E. Berti, Padova 1981, 263ff.

BIPARTITE SCIENCE IN ARISTOTLE'S BIOLOGY

339

an examination of the facts is necessary. One has to find out which facts can serve as proofs for other facts. The evident facts which cannot be proved must help to prove other facts while they themselves must be recognized as principles in a methodologically different way. Good parallels to this presentation of the relationship between principle and fact are E.N. I 2.1095 b 6 (&px� y&p To 8n) and I 7.1098 b 2 (To 3' 8n "PwTo" xoct &px-fi) . The facts are a poliori called principles. . another passage which is relevant in this connection. In There 1s Anal. post. I 23.84 b 22ff. Aristotle says: et 3< µ� foTt" (sc. Tt µfoo"), o6xon itcr"t'tv &:7t63e:L�Lc;, &:AA' � E:7tl "t'0Cc; &:pzOCc; 030c; ix0"t'1J Ecr"t'lv. " If there is no further middle term, no further demonstration is possible, but the proper method is the way to the principles."' As far as the search for middle terms is concerned, Aristotle does not in this context think of the apodeixis which runs from the principles to derived propositions, but he rather has in mind the proceeding questions of the researcher who is seaking for more and more explanations (i.e. logically for more and more m1 ddle terms) and so finally arrives at the principle. Only the end of _ this procedure and, in addition to it, the recognition of the principles on the ' 'way to the principles" , do make the apodeixis (in the strict sense) _ possible, the apodeixis, which runs in the direction opposite to the procedure here described by Aristotle. Apparently this procedure is an empmcal one. In terms of syllogistics it is a chain of syllogisms of fact. In the following paragraph the metaphor of intervals on a line is used (84 b 3l ff.). Aristotle speaks of dividing an interval BA (A being the principle) by further middle terms until we reach an indivisible, unitary proposition which no longer admits of further packin g.9 In connection with Anal. post. I 30 this passage becomes _ _ mtelhg1ble. The mere facts, which have been collected by way of easily icr"t'oplix and which are represented by propositions in which certain properties are predicated of a substance, these facts are only the raw material for the search for the oclTloct. They must be separated into defimt10nal and non-definitional propositions and brought into a logical connect10n. Before deduction can start (A is C because of B; A is D because of C), we have to ask: Why is D A? Why is CA? This can be made explicit in the form of a syllogism of fact, but can also be . graphically represented by an interval on a line marked by middle "terms". The principle itself cannot be elucidated by middle terms, but must be made evident by means of e7tocywy-lj and "ouc;. This means that two methods belong to the ascent: The "way to the principles" in

(8) Cf. the interpretation in M. Mignucci, L'argomenlazione dimoslraiiua in Arislolele. Cornmento agli Analilici secondi I, Padova 1975, 506f. (9) Cf. J . 1-1. Lesher, "The Meaning of NOY:E in the Posterior Analytics", Phronesis 18, 1973, 56 f.

1

340

BIPARTITE SCIENCE IN ARISTOTLE'S BIOLOGY

W. KULLMANN

'' \ .'' )

the strict· sense (E:nocywyfi, \lotii:;) and the scientist's continuous search, via middle terms, which aims at the definition from which deduction starts. We go on to the biologica l writings themselves.1° The distinction between <pocLv6µevoc and ochlocL occurs in PA I l .639 b 7ff., 640 a 14f. In IA 704 b 9f. the division 5TL/8L6TL is found. Apparently the HA provides the 5n, IA, PA and GA the 8L6n. This is confirmed by HA 1 6. On the one hand there are mentioned oci U7ti:Xpx_oucroct 8toccpopocl x.ocl TOC auµOe:0't}x6Toc 7t&at as well as lcrToploc, on the other hand ochlocL and oc7t68eL�L,. The distinction also recurs in the expressions E:� ©v and 7tepl &v. The first one signifies the premisses, the last one the conclusions." In which sense does HA provide the principles of proof in biology? In which sense are IA , PA and GA assigned to the presentation of the "why" and the causes? We begin with the first question . The introduction of HA is very much in keeping with the model of apodeixis as developed in the A nalytics. According to Anal. pr. I 30 it is necessary for the scientist to begin with collecting all propositions which could serve as premisses and conclusions in an apodeict ic syllogism . To this Aristotle's statement in HA I 6.491 a 9f. correspon ds: . . lvoc 7tp&Tov TOL' u7tocpxoucroc' 8Loc<pop&, xoct TOL cruµ6&11x0Toc "ii"' 1'
•' 'I !i ' l·.. i '

(10) Cf. W. Kullmann (see above n. I) 255ff. (11) Cf. W. Kullmann (see above n. 1) 81 and n . 8. (12) ll. Bonitz, Arislolelische Studien V, Sitzungsberichte der Osterreichischen Akademie der Wissenschaften, Wien, philos.-hist. Cl., 55, 1867, 13ff., esp. 39 (reprinted in: H.B., Arislolelische Sludien. Fiinf Teile in einem Band, Hildesheim 1969, 31 7ff.); Id., Index Aristotelicus 713 b 43ff.

According to this theory the term 8Loc<popocl in HA I 6.491 a 10 must be t.aken to mean only parts of specific differences. As regards the anonymous "greatest genus" of the reptiles and amphibia, such partial defimt10nal features are probably nTpa7to8oc, lfoo8oc, �cpoT6xoc, cjioT6xoc (and perhaps also q>o1'L8cuToc). Accordingly, the cruµ6e611x0Toc (scil. xoc6' ocuT
.

I ·. it..;: , I' 1:'.. 1:. i

1· '· '

I

i ', I

!1"i

li\,'i.

I., I· ' · \' i I .

!:

1 1!

I ,, 1 :

i: ' · ''

1 1.

341

But the written presentation of E:7tocywy� and of cruAAoytcrµol Toti 0Tt is lacking. In other fields of natural science the situation is different. In celestial physics (De caelo I and I I) for instance we find a lot of syllogisms of fact.13 In this field it is partly necessary to infer the facts by analogical reasoning because they are not accessible to direct observation . Take for example the proximity of the planets and the spherical form of the moon in Anal. post. I 13. The following two passages are very instructive in this context: De cael. I I 3.286 a 4ff.:

"although we are trying to pursue our inqu1r1es at a distance, not so much at a distance in space, but rather because very few of their attributes (i.e. attributes of the celestial bodies) are perceptible." xocl7tep 7t6ppw6ev 7tetpwµEvotc; 7tote:1'cr6oct 't"fiv ��TlJO"tv, 7t6ppw 8' oUx. oOTw Tcj) T67tcp, 7tOAU ae µiiAAov T(i:i TWv cruµOe0Yjx6TWV oc0To1'c; 7tepl 7t&.µ7tocv OAlywv €x.etv cx.la61)crtv. (13) See P. Moraux, "La methode d 'Aristote dans I'etude du ciel" in: Arislole et Les ' problemes de methode, ed. S. Mansion, Louvain 1961, 173ff., esp. 180f. (= Aristote Du ciel Collection Bude, Paris 1965, cviff., esp. cxiii); W. Kullmann, "Zur wissensch � ftliche� Metho� e des Ar�stoteles" , in: Synusia, Feslschrift Schadewaldl, hrsg. v. H . Flashar und K . Gaiser, Pfulhngen 1965, 247ff., esp. 255ff. (reprinted in: Die Nalurphilosophie des . Arisloleles [= Wege der Forschung, Bd. 225], ed. by G . A . Seeck, Darmstadt 1975 301ff., ' esp. 3!6ff.).

I

1 I

342

'l

W. KULLMANN

I I 12.292 a 15ff. " although we have very little starting-points and are situated at so great a distance from the phenomena concerning them (i.e. the celestia l bodies)." xocl7te:p µLxpac; lfxov-rocc; &.c.popµ&c; xocl -rocrocU't'YjV &.7t6cr-r.xcrLv &.7t€xov't'occ; -rWv 7te:pl ocU-ra cruµOocLv6v't'wv. In the biological writings we do not hear anything of Aristotle's research or of his scientific considerations. We can only reconstruct the gap between HA and PA from the presentation of his results in PA and from his theory in the A nalylics. He had to start from the propositions stated in HA. In I I 1 7.507 a 34ff., for example, he deals with the horned animals: "In the first place, those quadrupeds and viviparous animals which are horned and have an incomplete set of teeth, have four such stomachs." 7tpfu't'OV µEv y&p -r&v -re:-rpoc7t68wv xocl �c.po-r6xwv Ocroc µ� ecr't'LV &.µc.p008ov-roc -r&v xe:poc-roc.p6pwv, -r€-r-rocpocc; lfxe:L -roUc; -roLoU-rouc; 7t6pouc;. Before beginning with apodeixis Aristotle had to ask why these ruminants which possess a plurality of stomachs do not have teeth in both jaws and why they are horned animals. In order to answer these questions he had to refer to the law of compensation, according to which an excessive flux of material to the horns prevents the set of teeth to become complete. In terms of the A nalytics this methodological procedure can be described as the combination of two syllogisms of fact, i.e. syllogisms which are established by immediate premisses, but not by the cause but by the more knowable of two convertible terms (I 13.78 a 26ff.). We are passing on to our second question: In which sense does PA contain the presentation of the causes? After having completed his "research" Aristotle is able to give causal explanations. As for the ruminants he argues in the following order (which is opposite to the supposed order of the research): I I I 2.663 b 3 1 11.: "At any rate, in the larger animals there is present a surplus of this corporeal and earthy matter, produced as a residue, and this N ature makes use of and turns to advantage to provide them with means of defence. That portion of it which by necessity courses upwards she allots to form teeth and tusks in some animals, and to form horns in others. And we can see from this why no horned animal has incisor teeth in both jaws, but only in the bottom jaw. Nature has taken away from the teeth to add to the horns; so that. the nourishment which would

BIPARTITE SCIEN CE IN ARIST OTLE 'S BIOLO GY

343 norm ally be supp lied to the upper teeth is here used to grow the horns. "14 (tr. Peck)

I I I 1 4.674 b 7ff.: " Every one of the horned animals (such as the sheep the ox the goat the dear, and the like) has several stomachs; and the urpose 'of them i� . . this. Smee the mouth 1s deficient m teeth, the service which it performs . . upon �he food is deficient; and so one stomach after another receiv , es the food. (tr. Peck) -r�U-rwv �· €xa�'t'OV �Ae:louc; lfxe:L XOLAlocc;, oiov 7tp60oc-rov �oUc; ocr� ifAocc.poc; xocl -r&AAa 't'OC 't'OLClU't'OC 't'W� �lp WV, 07t<.V� e�e:L3� 't'�c; epyoccrlocc; et..f..e:£7te:L 7te:pl 't'�V -rpoc.p�V � 1 'YJ' au cr oµoc-ro 1 �e:�-roupyLa c; 3Lix. 't'YjV €v8e:Lix.v -rWv 086v't'wv, � 't'Wv xoLAL&v €'t'€poc 7tp0c; ; � E:'t'e:pix.c; 3e:xoµe:v11 't''YJV 't'poc.p�v.

�

The underlying scheme may be supposed to be as follows: Horned (animals) have an incomplete set of teeth . Rum inants are horned. Rum inants have an incomplete set of teeth. (Animals) which have an mcom · p 1 ete set o f teeth, have several stomachs. Ruminants have an incomplete set of teeth. Ruminants have several stomachs. This schematization is no mere play, though we have to concede that Aristotle dehberately avoids any pedan try. It helps us to understand the mtentwns he has in mind in the biological writings Beyond that, the question arises what exactly Aristotle mea �s when he announces to deal with "caus es". Acco rding to his usua l

(14) This �aw of com �ensation is in a remote way analogous to the axioms mentioned . in the P�slerwr Analyl1cs. It is valid without being part of the course of de':1?nstrat1on. An �ttempt to discover other "general principles" in the biological wri�ings c�n b� faun? tn � - Gotthelf, " First principles in Aristotle's Parts of Animals", in: Ph1losoph1cal issues in Aristotle's biology, ed. by A. Gotthelf and J G Lennox Cambr1· dge 1987, 185!!. ·

·

•

I

�:

i

1i'

!

I

'

.

.

344

lished in an apodeictic doctrine only the form al cause can be estab return to the example to Yet, nt). syllogism (universality requireme teeth m rummants see.ms quoted above, the incomplete formati?n of the achs, and the formation to be the efficient cause for the plurahty of stom the incom pleteness of the of the horns seems to be the efficient cause for woul d have to introduce teeth. In order to solve this contradiction, one and to interpret it as a the notion of a generalized efficient cause For the species_ does not characteristic of the eidos, i.e. the species. , not gener�lized sense deve lop, whereas the efficient cause in the strict es operative only m becom belongs to the field of the contingent and lly inaccessible for tifica scien single natural processes so that it is Aristotle. But he does not say so. is the final causes of Besides, one of the main themes of PA I I-IV the organs), �s is and s tissue the the parts (i.e. the functions of . . of the word the final announced in PA I. Since in the strict sense II post. ngency (Ana l. cause " also belongs to the field of conti order n i alized final cause 1 1 .95 a 6ff.), one woul d have to think o f a gener But this 1s not stated . cause al form the with to be able to equate it caused many misundersclearly by Aristotle in PA either, and this has . . tandings of this treatise. thmg to do with the As I take it Aristotle's indefiniteness has some species) is. of a static fact that his c�ncept of eidos (i.e. form or e of an amm al, when it character. The eidos is defined by the appearanc a passage from the by rated is in an adult state. This can be illust Metaphysics:

' . !

i'.

BIPARTITE SCIENCE IN ARISTOTLE'S BIOLOGY

W. KULLMANN

'

'· t::

Mel. El 8.1050 a 2ff.: to generation and time "But here, too, it is clear, that also according in substance ' . because actuality is prior to potentialit y. It is also prior ance (as for mstance subst what is posterior in genesis is prior in form and m comparison with man a and an adult in comparison with a child not), and because other the , eidos semen. For the one already has the iple' i.e. an end princ a rds towa s . everything which is coming-to-be move is 1s for the genes the and iple, princ the (for the 'for the sake of which' is sake of the end)." (tr. Ross, with changes) oO't'w 7tpo't'E:poc 't'1ji:; Ouv&:µewi:; xoc-rOC &.f.).,' oUv xixt 't'IX6Tfl ye: 3-YjAov 0-rL fi e:vE:pye:toc xi:x.t O·n 't'oc T(j ye\IEcreL Ua't'EfllX y!vE<JLV xcd. xp6vov. &:AAOC µl'jv xoct oUalq; ye, 7tp&'t'OV µE:v 7toi:; cr7t€:pµa:To<.; -rO µE:v -r& e:iOe:t xal -tjj oUafq; 7tp6-re:poc (oiov &vYip 7tOCL00i:; xo:l. &v8pw t TO yty\l6µe:vov xcd . y�p -fi31J €xe:t �0 e:10oi:; TO 0' oU), xcd. 8't'L &7to:v E:7t' &.px.Yiv �ocOl?:e: i:;) yE:ve:cn fi . TE:Aoi:; (&:px.Yi y&p -rO o?l Eve:xet:, 't'Oi3 -rE:Aoui:; OE E\IEXIX ·

i• I::!·. : ii

I!

for the origin In Aristotle's conception of eidos no allowance is made 10n of its funct the and al anim idual indiv the and the development of rd appearance of the parts. They cannot be gathered from the outwa as abstract as the fully grown anim al. Aristo tle's notion of eidos is not

I'·. ' i ,_ :

! ·.•

I•

I1' i1:. !1

',,1'. .1. 1 11. 1.1

,.

' i

,I

345

notion of species in modern biology. Aristotle does not formulate: " In this species we find this feature in the embryonal stage and that feature in the adult stage" . It is only in the adult stage that the eidos is fully reached.15 Whoever is interested in the question whether Aristotle was an essentialist or not, has to take this problem into account.16 These difficulties induce Aristotle to present the apodeixis in a special way, which is described in PA I l .642 a 3lff.: Oe:tx't'E:ov 3' oO't'cu<;, oiov 0't'L ea't'L µE:v Ti &:vlX7t\IOYi -rou81 :x,&.pLv, -roU-ro 8E ylyvE't"OCt 8t<X -r&.8e e� OCv&.yxYj<;. � 8' &v&.yxl'j O't'E µEv lo' u't"t 't"OCU't"OC ECf't"t\I ou-rw<; EX,0\l't"OC XOCt 1tEq>UX6't"OC . I paraphrase the passage as follows: "The apodeixis must be made in the following way: ' Respiration', for example, takes place for the sake of refrigeration (or for the sake of rendering life possible). But it necessarily comes to be by the activity of the lungs. Here necessity sometimes means that, if refrigeration will be the goal, it is necessary to possess respiration, sometimes that to a certain species refrigeration or respiration respectively essentially and necessarily belong." \

>I

<

fl

>I

I

According to this passage, as I take it,17 the final cause as well as the causation "of necessity" are to be shown, one after the other; in the latter case the expression "necessity" is again to be understood in two ways. Aristotle obviously intends (I) to show, with reference to a class of animals, what the aim of respiration is in the concrete case (i.e. the final cause, e.g. refrigeration), and (2a) to state that respiration and its efficient cause (! 3,& T1'3•) are hypothetically necessary to bring about refrigeration and render life possible in the concrete individual case; at the same time, it seems that he wants to maintain (2b) that refrigeration, the possession of lungs, etc., are essential properties of a certain class or genus of animals and therefore are absolute necessary. This interpreta­ tion goes back to William Ogle.18 It alone seems to allow a satisfactory explanation of the subdivision made shortly before in 642 a 31!. of the concept of necessity. It is not possible to assume that "accidental necessity" is meant, as had been suggested some time by David

(15) �or the term eido� as � enoti�g the visible aspect of a thing see now P. Pellegrin, _ " Logical difference and b1olog1cal difference: the unity of Aristotle's thought", in: Philosophical issues in A ristotle's biology (see above n. 14), 322. �16) Cf. D: M. B alme, "Aristotle' s Biology was not Essentialist", Archiv fiir _ 62, 1 980, lff.; revised version in: Philosophical issues in Aristotle's Gesch1chte der Phl/osoph1e biology (see above n. 14}, 291ff. (17) Cf. W. Kullmann, "Notwendigkeit in der Natur bei Aristoteles", in: Aristoteles.

Werk und Wirkung, Paul Moraux gewidmel, I. Arisloleles und seine Schule, hrsg. v. J. Wiesner, Berlin-New York 1 985, 207ff., esp. 218f. (18� The Works of Aristotle, translated into English under the editorship of J . A . Smith and W. D . Ross, vol. V De parlibus animalium, by W. Ogle, Oxford 11912: "such necessity as that which connects substances and their inherent properties and characters".

:,f ,,

346

W.

KULLMANN

Balme." This necessity derives from material and does not contribute to the aim, although it only exists under the condition that an aim be accomplished (e.g. the blue colour of the eyes according to GA V 1 .778 a 32ff.)." Balme rightly calls properties which are accidentally necessary (commonly Aristotle says: cruµooo(ve' ·� &v
,,

'

{19) D. M. Balme, Aristotle's De Partibus animalium I and De generatione animalium /, translated with notes, Oxford 1972, 101 and 80. , {20) Cf. D. Balme, "Greek Science and Mechanism I. Aristotle on Nature and Chance", ClQ 33, 1939, 135; W. Kullmann, "Notwendigkeit" (see above n. 17) 226ff. {21) Commentary {see above n. 1 9), 81. (22) This view is held by J. M. Cooper, Hypothetical Necessity, in: Al'istolle on Nalure and Living Things, Philosophical and flistorical Studies, presented to D. M. Balme,

i' ' 'I iI' Ii ul , .

'

' 1: ' ;. 1 t

',,.i I, 1

,1 , 1 \� i

ed. by A. Gotthelf, Pittsburgh-Bristol 1 985, 155; Id., "Hypothetical necessity and natural teleology'', in: Philosophical issues in Aristotle's biology {see above n. 14), 258rf.; D. M. Ba\me, "Teleology and Necessity", ibid. 285 and n. 33 {converting from his former vie\V, see above n. 1 9). {23) D. Frede, "Aristotle on the Limits of Determinism: Accidental Causes in Metaphysics E 3", in: Aristotle on Nalure and Living Things {see above n. 22) 215ff. Additions to the notion of necessity like "taken by itself", ibid. 216f., are qualifications which are incompatible with the status of absolute necessity. {24) This chapter is not taken into consideration by the authors quoted in n. 22 and 23. - See also W. Kullmann, Wissenschaft und Methode {see above n. I), 21ff.

347

BIPARTITE SCIENCE IN ARISTOTLE'S BIOLOGY

But the passage is much disputed, and will remain difficult." Therefore I leave it open whether (apart from Anal. post. I I 1 1 and the beginning of PA 126) there is an attempt of Aristotle to approximate final cause to formal cause. But I have no doubt that in PA he virtually thinks of a generalized final as well as efficient cause. · Nevertheless he is unable to conceive the species in such an abstract way as we do. In comparison with other stages in the development of an animal a certain stage is distinguished by its perfection. Therefore Aristotle necessarily shrinks back from identifying concept­ ually the efficient and the final cause with the formal cause. These problems point back to the Platonic background of Aristotelian science and have only had a small influence on the later reception of Aristotle's work. Theophrastus adopts the bipartition of s�1ence in Lcr't'opla./a.i't'L<XL for botany, but with one important difference. The treatise De causis plantarum is confined to the treatment o f the eWcient cause. The final cause, which is of prime . importance m PA , 1s touched upon by Theophrastus only in his Hisloria planlarum, and even there but occasionally.27 The reason for that is not that Theophrastus attributed less importance to the final cause than Aristotle did. The explanation is rather to be found in the different subject of the botanical science. Theophrastus in fact follows Aristotle in his estimation of botany: as far as plants are concerned the final cause has a smaller role than in zoology. In Alexandrian medicine (in the 3rd century B.C.), which was to a large extent influenced by Aristotle, much emphasis was given to the cognition of the aMoc,, in opposition to the younger empiricists. Erasistratus, a pupil of Strata, wrote a treatise 7tepl octn&v. He was probably dealing ex•lusively with the efficient cause. Thereby he answered those who claimed that diseases are &vocmoMY'll""' (fr. 25 Deichgraber)." By the mediation of Galen as well as later on by direct translations of Aristotle the model of bipartite SC!ence was transmitted to the methodological doctrines of mediaeval and modern philosophy, to Grosseteste, to Roger Bacon, to Galileo and to many others.29*

Kullmann, "Der wissenschaftliche Charakter der Biologie des . (25) But . cf . O.w. Ar1stoteles. E1ne berprOfung", in: Gedankenzeichen. Fesischrifl Klaus Oehler, ed. by R . Claussen and R. Daube-Schackat, Tiibingen 1988, 16ff. (26) Cf. W. Kullmann, Wissenschaff und Melhode (see above n. 1) 277ff., 13ff., Id . , "Der wissenschaftliche Charakter" {see above n. 25) 14ff. (2?} Cf. G: WOh �le, Theophrasls Methode in seinen bolanischen Schrifien ( Studien zur anl1ken Phiiosophie, hrsg. v. H. Flashar, H. GOrgemanns, W. Kullmann, Bd. 13), Amsterdam 1985, 84ff. {28) Cf. K. Deichgraber, Die griechische Empirikerschule . Sammlung der Fragmente und Darslellung der Lehre, Berlin 1930, 278f. (29) Cf. A. C. Crombie, Robert Grossetesle and the Origins of Experimental Science 11001700, Oxford 1953. * I thank my son, Dr. Thomas Kullmann, for his help in establishing the English version of this paper. ·

·

=

BIOUXllE, l.OG!QU.I! £l' MIITAPHYSIQUE

chez ARISTOT£

8eminaire CNRS-N.S.F., 1987

Editions du CNRS, Paris, 1990.

NECESSITY AND EXPLANATION IN ARISTOTLE'S ANAL YTICS ARYEH KOSMAN

In this paper I propose to consider two questions about necessity. The first concerns a problem in the interpretation of Aristotle's modal logic. In Book I of the Prior A nalytics, in the course of discussing modal syllogisms, Aristotle makes claims that seem inconsistent with one another, or that seem based on inconsistent views about the nature of modal propositions; is there any way to make Aristotle's modal logic coherent? The second is a question about the nature and role of modality in the task of scientific explanation. In Book I of the Posterior A nalytics, Aristotle argues that episteme, the understanding that comes about by explanation and enables explana­ tion, is of the necessary and proceeds from the necessary; what does he mean by this? It is not obvious that these two questions are related to one another, but I hope to show that they are. I will suggest that it is fruitful to think of the second question as involving the interpretation of that concept of necessity whose purely formal consideration in the Prior Analytics generates the first question. The scope of this essay is thus a modest one; it is to suggest that well known difficulties in Aristotle's modal logic may become clearer when we recognize the unity of the enterprises which shape the discussions of the two halves of Aristotle's A nalytics. I . Aristotle's account of mixed modal syllogisms. First, then, about modal syllogistic; the story I am about to tell is by now a familiar one.1 Partway through Book I of the Prior (1) It is discussed by A. Becker, Die arislolelische Theorie der MOglichkeitsschliisse (Berlin, 1933), by 'stoors MacCall, Aristotle's modal syllogisms (Amsterdam, 1963) by

I

i I

t' I 1.1 11 1:1.:.·,°1 '!i'. !I.

ill :

350

A. KOSMAN

Analylics, Aristotle considers what we might call hybrid modal syllo­ gisms, and in particular, in Chapter 9, he considers syllogisms one of whose premises is necessary and the other of which is assertoric. The outcome of such syllogisms, he argues, differs depending upon which of the premises exhibits which of the modalities. For if, he claims, the major premise of a demonstrative syllogism is necessary, there follows a conclusion that is also necessary; but if it is the minor premise that is necessary, the conclusion that is generated is merely assertor)c. "For example," Aristotle writes, assume that A belongs or does not belong to B of necessity, and that B belongs simply to C. If we make this assumption, then A will also belong or not belong to C of necessity. For since A belongs or does not belong of necessity to every B, and C is one of the Bs, it is clear that A will also of necessity either belong or not belong to C. But if the AB premise is not necessary whereas the BC premise is necessary, the conclusion will not be necessary. For if it were, it would follow that A belongs to some B of necessity, by the first and third figure. But this is not true.2

Consider, for the sake of simplicity, only cases of belonging, that is, consider syllogisms consisting of universal affirmative premises. Aristotle's claim is that in the case of these two hybrid Barbaras, the inheritance of the trait of necessity follows the modality of the dominant major premise rather than that of the recessive minor premise. Necessarily AaB and BaC yields necessarily AaC, whereas AaB and necessarily BaC yields merely AaC.3 This claim of Aristotle's has constituted for many commentators a difficult crux in Aristotle's modal logic. According to Alexander, the claim met with immediate disagreement on the part of Aristotle's successors; in his commentary on the Prior Analylics, Alexander explains Aristotle's remarks, and then adds: But his followers who were in the circle of Theophrastus and Eudemu� deny this; they say instead that in all cases in which a combination' of a necessary and an assertoric premise constitute a syllogism, the conclusion that results is assertoric. They infer this from the principle that in all such combinations the conclusion is similar to the least and wea ker premise.4 _

Nicholas Rescher, "Aristotle's theory of modal syllogisms and its interpretation", in The ed. Mario Bunge (Glencoe, 1964), 152-177, by Wolfgang \Vieland, "Die aristotelische Theorie der Syllogismen mit modal gemischten Priimissen" , Phronesis 20 (1975) 77-92 and most recently by Richard Patterson, "The case or the two barbaras: Basic approaches to Aristotle's modal logic", forthcoming in Oxford

critical approach to science and philosophy,

Studies in Ancient Philosophy. (2) Prior Analylics I, 9, 30a 1 5 ff.

(3) Where "PaS" symbolizes "P belongs to all S", "PoS" "P belongs to no S", etc. (4) Alexander Aphrodesiensis, In Arislotelis Analyticorum Priorum Librum I Commentarium, ed. M . Wallies (Commentaria in Aristotelem Graeca 2.1 , Berlin, 1883) 124,8ff. Cf also 173,32ff. and 1 19,? ff.; John Philoponus, Jn Aristolelis Analytica Priora commenlaria, ed. M. Wal\ies (Commentaria in Aristolelem Graeca 13.2, Berlin, 1905)

NECESSITY �ND EXPLANATION IN ARISTOTLE'S

ANAL YTICS

351

The " least and weake r" principle to which Aristotle's critics appea l, and :"h1ch was to becom e the Scholastic modal dictum peiorem semper sequzfo r conclusw parlem, reveals a more Mende lian model of modal mhentance at work m the thinking of Theophrastus and Eudem us; on their understandmg, 1t 1s the trait of necessity itself, rather than one or another of the parent premises, that is recessive, and the trait of non­ . necess1ty that 1s domin ant. 2. Whal this account suggests aboul Arislolle's lheory of modal proposi­ tions. We might interpret the disagreement between Aristo tle and Theophrastus by supposing that the two have different pictur es of the deep structure of modal propositions. Suppose that Theop hrastus . . (msp1red perhaps by a readm g of On Inlerprelalion) thought of modality as a feature of the entire quanti. fied proposition, and not as a feature either of the proposition prior to quantification or of one of its argu ments. Thinking the modal operator thus located, as it were, . outside the quantified premise and thinking of the syllogi sm on the mode I of a cham only as strong as its weakest link, he then disallowed . the � nfere nce of modally "stronger" conclusions from "weaker" premises. ....

Aristotle's view, on the other hand, suggests an interpretatio n of th e structure of necessary propositions that places the modal operator . wilhtn the scope of the quantifier; universal necessary affirmative . propos1twns, for examp le, have the form (x)[Sx--+ NecPx ]. Such a conception generates precisely the distinction Aristotle makes . For it reveals one of the hybrid Barbaras to lead to a necessary conclusion: (x)[Mx --+ NecPx] and (x)[Sx --+ Mx] yield (x)[Sx --+ NecPx]. But it re­ veals the conclusion of the other Barbara to be restricted, just as our . text reqmres: (x)[Mx --+ Px] and (x)[Sx --+ N ecMx] clearly do not yield the necessary concluswn (x)[Sx --+ N ecPx]. We might be concer ned that (x)[Mx --+ Px] and (x)[Sx --+ NecMx] do not immed iately yield any concluswn; but on the assumption that N ecp --+ p is a valid modal . prmc1ple, we can easily generate a more fruitful minor : (x)[Mx --+ Px] and (x)[Nec Mx --+ Mx] yie ld (x)[N ecM x --+ Px]. And that minor , together . . with the or1gmal maJOr, will yield JUSt the assertoric conclusion which Aristotle claims follow s from this syllogism. In addition to such formal considerations, this reading may be . attra�t1ve for other reasons. For it may suggest that Aristotle's doctrme of the two hybrid Barbaras demands an interpretation of

�;���;lartum,

205,13 ff.; �23,15ff.; 129,16 f�.; Ammonius, I n Arislolelis Analyiicorum Priorum fibrum ] ed. M. Walhes (Commentaria in Arislotelem Graeca 4.6, Berlin, 1899)

A . KOSMAN

352

i'

1. ··'.1.: ., ' ,, 1:

'

' ,1 :

modality and of the modal proposition that is de re rather than de dicto, _ _ suggesting as it does that necessity holds of mdividuals mde pendently of _ the description under which they are mtroduce d. And this may agree _ a feature of the _ with what we expect from Aristotle; necessity is belonging of a predicate to a subject, and is not to be understood m terms of analyticity.' 3. A difficulty in Aristolle's theory. The story might end here, with what seems a plausible account of his understanding of the structure of modal propositions, were it not for Aristotle's repeated claims that modal propositions convert m the same _ way as do simple assertoric propositions. This is how Aristotle opens chapter 3 of Prior A nalylics I : .

it works the same way with necessary premises. For � universal negative converts universally, while each of the affirmat1ves c?nverts kala meros. For if A necessarily belongs to no B, then � necessarily belongs � no A. For if it might belong to some, then A might belong to some

i

·

This is not simply an isolated claim on Aristotle's part; the conversion of modal propositions is a critical - we might even say . necessary - part of his general discussion of the modal syllogism . For . it is the conversion of modal propositions that allows for the reduction of modal syllogisms in the second and third figure to the perfect syllogisms of the first, just as in the case of the standard assertoric syllogistic. The conversion is therefore no minor part of the modal syllogistic. But the account we have been considering, which locates the modal operator within the quantified proposition, will not allow. such conversion to take place. No obvious rule of logic could allow us to infer (Ex)[NecSx&PxJ from (x)[Sx -+ N ecPxJ or from (Ex)[Sx&N ecPx], nor to infer (x)[Px -+ Nec - Sx] from (x)[Sx -+ Nec - Px]. , . The conversions Aristotle allows could be understood if we w�re to imagine the modal operator as outside the entire quantified proposition, so that necessary A propositions, for example, have. the form Nee . (x)[Sx -+ Px]. But such an understanding makes � nexphcable the syllogisms of mixed modality, and suggest a relationship between

(5) This is a more complex and in many respects �eparate issue from t�e one w�th which we are here concerned; I have considered the q�es�1on of whether there is a. doctr�ne of de re necessity in Aristotle in a forthcoming piece ' Aristotle.1 s. essences and Anstotehan essentialism". The reading of the structure of modal propos1t1ons suggeste.d here seen:is also to be assumed by Aristotle later in his di�cuss1on of �odal r�ason1ng wh�n, in chapter 17 of Book I, he rejects the conversion of negative un1�e�sal contingent . Propositions I argued for this point in an earlier essay more optim1st�. c about .th e possibility of an easy coherence 1n Aristatle' s mod aI Iog1· c , "Aristotle on 1nconvertib1e modal propositions" Mind 79 (1970), 254£f. (6) Prior Analytics I 3, 25a28 ff. ·

I: ·[ii I: i1 .'. I II: ii I '

·

NECESSITY AND EXPLANATION IN ARISTOTLE'S

ANAL YTICS

353

modality and the quantified proposition that we should not expect from Aristotle. 4 . A possible solution to the difficulty in Aristotle's theory. What should we do? We might follow the strategy of Aristotle's earliest followers and interpret the mixed Barbaras as a simple mistake on Aristotle's p art. Albrecht Becker, in a monograph published in 1 933, offers another suggestion.' In Becker's view, there are two different Aristotelian interpretations of the deep structure of the modal proposition, one at work when Aristotle considers the two Barbaras, and the other at work when he considers the conversion of modal propositions. Becker rests much on Aristotle's claim in chapter 13 of Prior A nalylics I , when, in a parallel discussion of contingent propositions, he observes: The expression it is possible for this to belong to that may be understood in two ways: [it is possible for this to belong] either to the thing to which that belongs or to the thing to which that can belong.8

This observation seems to Becker to yield two different Aristotelian interpretations of what I have called the deep structure of modal propositions, in this case of propositions asserting possibility: a contingent A proposition may be thought of as a proposition (!) with the structure (x)[Sx -+ PosPx] or (II) with the structure (x)[PosSx -+ PosPx]. If we understand these two structures as governing Aristotle's discussion of modal propositions in general, then the anomalies which we have noted in the modal syllogistic are explicable. For as we have seen, construing necessary A propositions as having the form (!) (x)[Sx -+ N ecPx] makes understandable the distinction between the two forms of syllogism with premisses of mixed modality. If, on the other hand, we think of such propositions as having the form (II) (x)[NecSx-+ NecPx] and of the corresponding I propositions as having the form (Ex)[NecSx&NecPx], Aristotle's remarks concerning conversion become understandable, and without the necessity of imagining the modal operator outside the entire quantified proposition. Becker's interpretation, however, merely uncovers the source of the anomaly in Aristotle's thinking; on his interpretation, Aristotle's modal logic remains muddled, resting on two quite different understandings of the structure of modal propositions. But it is difficult to imagine, if Becker is right, that Aristotle was simply confused in his sense of this distinction or of its application to the modal proposition. This is due not merely to the strong hermeneutical principle Arisloleles raro confusus

·

(7) Becker, Die arislolelische Theorie der M6glichkeilsschliisse, op. cil.

(8)

Prior Ana(ylics l,

13 32b15.

354

A. KOSMAN

est; in this case that hermeneutical principle is bolstered by the very statement of Aristotle on which Becker rests his distinction, as well as by other evidence of Aristotle's understanding of the . distinction.' Becker in fact does not accuse Aristotle of confuswn, but merely of ambivalence. But the question we need to ask if this ambivalence is not to be read as a sign of confusion is this: is there any way in which we can understand the ambivalence to be grounded in a . coherent project of Aristotle's? Nicholas Rescher has more recently offered an alternative unders­ tanding of the mixed modal syllogisms which is ingenious, but whose baroque complexity is difficult to discover in Aristotle's discussion.10 Rescher's interpretation, however, has the considerable virtue of interpreting the modal syllogistic in light of Aristotle's discussion in the Posterior Analylics. It thus reminds us that the discussions in the two parts of the Analytics are of a piece, and that the semantic interpretation of the Prior Analytics' formal notion of modality may most aptly be looked for in the necessity which the Posterior Analylics deems requisite for scientific understanding. It is for this reason that I turn to the first book of the Posterior A nalytics and to the question of explanatory modality.

·. 'I

5. Understanding and necessity. In the Posterior Analytics the recognition of necessity is said by Aristotle to be a mark of proper scientific understanding. We think we understand something in an unqualified sense, and not in · t�e sophistical or accidental sense, when w� know both of the cause that 1s _ _ 1s the cause and that things responsible for its being the case that 1t could not be otherwise. 11

If we think of Aristotle's discussion of episteme on the model of modern discussions of knowledge, in a context governed by concepts of doubt, justification and certainty, it may seem to us that the demand that the proper object of episteme be necessary is meant as a cond1twn for the possibility of certainty, as a guard against the .eventuality of error. But once we have come to recognize that the governing contexts of the Posterior A nalytics are those of understanding and explanation !' a

.I I:

(9) In Sophistical Refutations 166a23, for example, where the scholastic distinction between modality in sensu diviso and in sensu composilo seems to have its origin. (10) Nicholas Rescher, "Aristotle's theory of modal syllogisms and its interpre­ tation'', op. cii. (11} Posterior A nalytics I 2, 71b9 ff. , (12) As I hope we by now have come to recognize: see my "Understanding, explanation and insight in Aristotle's Posterior A nalylics", in Exegisis and argument (edd. E. N. Lee, A. P. D. Mourelatos, R. M. Rorty), Phronesis supp. vol. I (Assen, 1 973) and M. F. Burnyeat, "Aristotle on understanding knowledge", in Aristotle on science. The Posterior Analytics (Proceedings of the Eighth Symposium Arislotelicum, ed. Enrico Berti, Padua, 1981).

NECESSITY AND EXPLANATION IN ARISTOTLE'S

ANAL YTICS

355 somewhat different interpretation of Aristotle's remar ks will seem appropriate. We say we have understood something when we think we know (and therefore can reveal) why it must be the case that it is so. There is not yet understanding of such a thing being the case as long as we have only shown that it might be the case; for we could still ask; why, given _ doesn that 1t 't have to be the case, is it the case? That question, which _ ely the dema is precis nd for explanation, will be answered only when we recogmze that given these and these causes, it must result that such and such is the case. The explanatory dema nd, in other words, is fully _ satisfied and broug ht to rest only by the invocation of necessity. Aristotle's requirement of necessity in scientific under standing does not, therefore, reflect a theory according to which , for the sake of _ certain ty, only the necessary can be the object of prope r knowledge, but a theory according to which necessity is a feature of explanation, since successfully to explain something is to show why it must be so. The necessity in question 1 therefore, is a necessity which comes to be revealed in the development of expla natory understanding; it is not a necessity of which we need be aware prior to and independen tly of scientific d emonstratio n. Nor is it, therefore, the necess ity of a pragma _ s1mph c1ter, but of a conclusion insofar as it has been explained, that is, revealed to be necessitated by the cause invoked in the apodeixis which explains it. The nature of explan atory modal ity is made clear when Aristotle subsequently observes that the necessity of what is explai ned must be revealed by way of explanation from premises which are in turn ? ecessary in just the . same sense. I n one sense 1 this further requirement is a consequence of the fact that, as the recursive nature of Aristotle's theory makes clear, there is not yet scientific unders tanding of . something if that by virtue of which it is explained is not itself understood. But in additi on, it makes clear that no necess ity of the conclusion simpliciter is sufficient to satisfy Aristotle's theoretical deman ds. In the sixth chapter of Posterior Analytics I, Aristotle stresses that even when a conclusion is necessary, there will not be full explanatory understanding unless both of the premises as well exhibit the nec�ssity in ques tion; if either premise fails to reveal a necessary _ connectwn of the middle , there will not be an understanding of the reason why the conclusion must exhibit the connection it does and there ' will therefore not be episleme properly understood.'' Consider more generally what's deman ded in an explanation. A certain pragma consists of one sort of thing being another; the office of

{13)

Postertor Analytics

I 6, 74b26 ff., 75al ff.

·,' ,

l .\ '

I

I '.I

356 .

A. KOSMAN

explanation is one of revealing why this has to be so. An explanation, in other words, attempts to show why an entity that may be identified under one description may, just because it can be identified under that description, also be identified under another description. This possibili­ ty of redescription is revealed to be a feature of the necessity of connection among alternative descriptions; an explanation attempts to show in other words, that it is not simply an accident that some entity which is F is also G, since the description by which the entity i� specified as being F is connected by some sort of necessary connection to the description by which the entity is specified as being G. This explanatory office, as we know, is accomplished by the middl e. . . This means that the explanation we want is one in which both premises are necessary. For if we want to show the n �cessary . connection of two descriptions by means of another description (the middle), then for each of the first two descriptions, the connection of that description to the middle must be understood, and must the refore . exhibit the same kind of necessary connection that we are trying to achieve for the conclusion, that is, for the explanandum that is to be the immediate object of our understanding. Each extreme term, therefore, . (that is each of the descriptions which determine the poles of predicative being of the explanandum) must be connected to the middle by necessity if the middle is to serve as an explanans. Jn the paradigm of scientific understanding, therefore, explanation is from necessary premises; for we have a phenomenon under a �erta1 n . description which brings together two terms, and explanation is achieved through the middle which, by specifying the phenomenon under a more general mediating description, reveals the expbnator.y . connection between the two terms. Each of the premises by which this explanatory mediation is accomplished is standardly itself a conclusion of some prior apodictic syllogism. And since the conclusion . of a . syllogism is simply the perfected form of the premise s of which the . syllogism consists,14 the necessary premises of a scientific explanation reveal a connection effected by prior necessary connectiol}S to a prior middle which has . been sublimed in the subsequent act of inference. Conversely, this middle is invisibly present in our prescien­ tific experience and is only brought out in those acts of epagoge and . apodeixis by which the scientist reveals that middle which explains the connection. And this is for Aristotle precisely the essential pro1ect of science; science just is the act of revealing the ramified syllogistic of what in the first moment of scientific awareness appears to us as conclusion, without it yet being understood: conclusion of what.

{14} This is why "symperasma" sometimes means the terms of the syllogism, as at I I 1 , 53a 17, where Waitz and Bonitz introduce a derivative sense.

Prior Analylics

I I

NECESSITY AND EXPLANATION IN ARISTOTLE'S

ANAL YTICS

357

6. Syllogistic necessity and the necessity of scien tific understanding. We turned to the question of necessity in the Posterior Analylics following the suggestion that the so-called Prior and Posterior A nalylics form a single work, whose subject matter is, as Aristotle announces in its opening lines, explanation and expla natory understanding: apodeixis and episteme apodeiklike.15 I f this suggestion is correct, then we shoul d expect the issues of the first two book s of this work to be preparatory to the discussion of episleme apodeiklike in the latter two books · the structure of syllogism, for exam ple, is investigat ed because apodei� is is accomplished by syllogismos, that is to say, becau se scientific expla na­ . tion takes place by mean s of reaso ned discourse. This is not to say that the formal logic of the Prior A nalylics is nothing but the hand maid en of the subsequen t account of scientific explanation in the Posterior Analylics. Here, as standardly in Aristotle the �hape of discussion owes as much to Aristotle's natural curiosity, hi� passion for completeness, and his associative style of organization as to any clear agenda of premises in a large architectonically ordered argument. The formal structure of syllogistic clearly takes on a life of its own in the Prior Analylics, and Aristotle's inter est in the elaboration of a formal logic of deduction reaches well beyo nd his imme diate needs for the a �gument concerning apodeixis in the Posterior A nalylics. Nor is it to say that the Analylics is a manu al for carrying out the . ities activ of explanatory science. The Analylics is an abstract and form al theoretical essay ; it presents a mode l of scientific understanding by ideal izing and formalizing the structures of normal expbinatory . ity, activ i.e. o f science. It is not, so to speak, a theory of a form ally . abstract and ideali zed scientific practice, but a form alized theory of . . actual scientific practice. Scientists do not const ruct syllogisms in one figure or another of the sort we encounter in the Prior Analylics; they construct complex bodies of explanatory discourse, such as we find, for exam ple, in the Paris of Animals, and we are able to understand these bodies of discourse as instances of formal syllog istic only when the philosopher has performed his (characteristically posterior) analysis.16 But given these qualifications, we shoul d expe ct to find that the . choic e and treatment of topics in the two works is in general governed by a single interest, the interest in the nature of explanatory science, . and its most obvious activity, that of scientific expla nation. And so we

(15) Prior Analytics I 1 , 24a10 . (16) See the extended use of this principle by Allan Gotthelf in his essav "First principles in Aristotle's Paris of Animals", in Philosophica l issues in Arislolle's biol�gy (edd. A. Gotthelf and J. ,Lennox Cambridge, 1987), I90ff.

358

A. KOSMAN

should expect to discover that the necessity of Analylics I, that is, of the first book of the Prior A nalylics, is the formal analogue of the necessity _ of scientific explanation demanded in A nalylics I l l , that 1s, m the first book of the Posterior Analylics. . . the Standardly, therefore, a modal logic which purports to exh1b1t paradigmatic form of explanatory reasoning will be concerned with propositions which are of the form Necessarily PaS - P belongs necessarily to all S - and whose deep structure may be represented as (x)[NecSx � NecPx]. For each of �uch propositions will represent a sublimed form of two necessary premises pr10r to it; each will represent a conclusion in which the prior necessary linking middle which mediates those premises has been suppressed. Propositions of this form, which clearly convert, constitute the standard type of propos1t10n that Aristotle has in mind and exhibit the standard structure of modal necessity invoked throughout the Analylics. It is this structure which Becker properly recognizes in his second model of the modal proposi­ tions a structure that explains Aristotle's standard modal syllogistic and h is standard forms of modal conversion. But is there anything in this account that might help us understand why Aristotle makes the distinction between the two forms of mixed _ _ which struck us as modal syllogism with which we began, the d1stmct10n recommending Becker's first model of modal structure, a model . according to which one thing is said to belong necessarily to subjects to which another belongs simply? For this we need to thmk agam about the question of explanatory necessity. 7. Necessity and explanation again. Our aim in the discursive activity of science is to understand why some s or other is some P or other. This understandin� , :'e sa,w, is achieved by apodeixis, by an explanation that shows why 1t 1s the· case _ that an entity under a certain description must also be spec1fiable under · _ another. Standardly, I have suggested , such explanat10n occurs m contexts in which the two descriptions are related necessarily through a third description which constitutes a middle, in turn related necessarily to each. S has to be P because S has to be M and M has to be P. _ But if science has both its source and telos in the part1cular1ty of experience, we will wish to be able to countenance explanations of a different sort as well. For we will want to be able to account for instances of reasoning by which an entity under a certam descr1pt10n comes to be understood to exhibit another being, even though the first description may not play a critical role in effocting the u� derstandmg; _ thus to that description may merely pick out the entity who se bemg 1s _ _ be explained, but not be related necessarily to the rruddle that explams why it exhibits the being it does.

NECESSITY ,AND EXPLANATION IN ARIS TOTL E'S ANALYTIC$ 359 Consider for example these cases. We wish to understand why the moon IS m echpse. We are offered this explanation: the moon is (at the moment) a body with an opaq ue substanc e interposed between it and its light source, and beings of that sort (for these and these reasons) _ _ _ echpse. Or 1mag me our w1shmg to understand why the interior angles of a certam figure are equal to two right angles. We're given this explanat10n: the figure m question is a trian gle, and the interior angles of a triangle (for these and these reasons) are equal to two right angles. It seems that m each of these cases we have been given an apodeixis the formal repr�sentation of whic h is considere d in the A nalylics; for have n't we been given m each case an expl anat ion whic h is productive of scientific understanding? But in each of these cases, the entity in ques tion is initially specified _ by a descr1pt1_ 0n which 1s not necessarily connected to the explanat ory midd le. In order to reahze that fact, cont rast the mini mal explanations . given with the deeper explanations pare nthetically figured in each case by the phrase "for these and these reasons" . The preliminary midd les body with an opaque substance interposed between ii and its light source and triangle � are related by necessity to these further and deeper explanatory midd les whwh relate them necessarily to the predicate in quest10n; but it is only contingently true that the moon is a body of that sort or that the figure in question is a trian gle. Each of the entities in question, therefore, exhi bits the character istic whic h is expl aine d, but does not exhibit that characteristic nece ssarily insofar as it is specified _ _ by its 1mti al descr1pt10n, but rather insofar as it can be specified by some other description to which the characteristic is related necessarily. It is therefore only in a qual ified sense that one is able to explain why the figure before me has two right angles, or why that shining globe abov e the horiz on eclipses. Wha t can really be explained, by mea ns of the further and more reve aling midd les, is why a triangle, whic h the figure before me happens to be, has two right trian gles, or why a body with an opaque substance interposed between ii and its light source, which the moo n happens to be, eclipses." In the case of these initial explanations, then , the connection to the explanatory middle expressed in the mino r premise is not a necessary one, although that of the majo r is. Whe n Aristotle considers the mod al -

(17� I think that the relation betwe n these initia . h, l explanations and the deeper � . expl�nat1ons whic as we will see, they to, bears an interesting resemblance to the �:t�tion between A-�ype and B-type explapoint nations that James Lennox develops in his essay _ an d expla D.1v1de in: . the Posterior A alylics in practice", . in Philosophical issues in Aristotle,s b1ology, op. cil. 90 ff. It may � d be the case that the entire discussion whic I've here devoted to explanations of �he indee particular qua individual is more appropriate withh . ns respe�t to explanatio of the particular qua what is kala meros in the sense Lennox explains. ,

' ;I I'

360

A. KOSMAN

chapter of status of the premises of scientific demonstration in the sixth _ trat1ons emons � that seen, have we as , denies Posterior A nalylics I, he tan­ whose premises are not necessary are capable of generating unders to out ts se ics Analyt ior Poster the which sense ding in the full _ he says, _ clarify. Among the arguments he sets out is the following; if,

[dia ii], even someone who doesn't have an account of the causal reason then if anding, underst ic scientif have doesn't tration demons when there's a [the ch \ .wh through B, term . A belongs necessarily to C, but the middle doesn_ t kn_o� h rily, necessa belong not � does trated, demons is ion] conclus . , s1n?e i t _s the reason why [dioli]. For it's not so by virtue of the middle conclus1on is possible for the middle not to be the case, whereas the ' necessary .18

For, Aristotle continues in an interesting and revealing aside,

through when the conclusion is necessary, nothing prevents the middle 19 ry. necessa being not from trated demons is which it

ering The preliminary modes of explanation we have been consid which to ns stratio demon ect are, I suggest, instances of just these imperf _ scientific Aristotle here denies the power to produce full formally understanding. N ote further that these demonstrations . are we which with ity modal mixed of sms represented by the very syllogi on ntary comme his m Barnes by ained began. This possibility is entert _ _ m his words, it, finds he e becaus sed dismis the Posterior Analylics, but t to "piqua nt . . . but perhaps too subtle" .-". But it seems to me difficul refers le Aristot which to forms stic syllogi the that sion avoid the conclu forms in this early chapter of the Posterior Analytics are precisely those ry necessa of mixed modal reasoning he discusses earlier; here is the of ty necessi some major, the merely assertoric minor, and that trouble 9. conclusion which we encounter in Prior A nalytics I I n these cases, scientific understanding in the full sense has indeed which not yet been accomp lished. · For the deeper causal explan ation tw o havmg its to e triangl qua necessarily links, for examp le, the figure _ d is reveale been has what But d. right angles has not yet been reveale the very nature to which we must look if we are to find s,uch a deeper explanation; what has been revealed is the fact that the figure's having as two right angles can only be understood insofar it is _apprenhended ified unqual of denial The figure. this as merely not this triangle, and and understanding is thus a sign of Aristotle's commitment to the powe_r of ise the necessity of proper and essential description in the enterpr only le Aristot scientific explanation. The object of understanding is for be qualifie dly the contingent and particular fact that is at first to

(18) Posterior Analytics I 6, 74b26 ff. (19) Posterior A nal!j lics I 6, 75al ff. (20) Jonathan Barnes, Aristotle's Posterior Analytics (Oxford, 1 975}, 125.

NECESSITY AND EXPLANATION IN ARISTOTLE'S

ANAL YTICS

361

explained, for example, that a certain figure has two right angles; properly it is that anything which falls under a certain description has two right angles. It is thus as we have seen only contingently true that this figure has two right angles; necessity attaches only to our understanding that a triangle, which this figure happens to be, has two right angles. But what follows from this is important; it is that the first essential task of scientific explanation will be the noetic task of correctly describing particulars, of finding the right universal to characterize those beings which have been picked out by the mere lhises and thats of our initial discourse. It is important that this is a task of proper description, and that understanding in these cases is not absent, but hidden, obscured by the distracting overdetermination of the particular. Our ultimate objective in the enterprise of understanding - whether as scientists or as philosophers - is not to turn away from the particular but to make it manifest, to demonstrate it. The aim is not to leave it in darkness while we attend to and shed the light of nous upon the universal and abstract; it is to reveal the particular by casting it in that light which shows it for what it is and why it is what it is. Our first task in that enterprise must therefore be one of discovering the particular in terms which do more than merely indicate it to us, but which disclose as well that nature which will be the subject of a more revealing apodeixis. The general form of such activity is epagoge, which Aristotle describes as the activity that "reveals the universal by making clear the particular" (deiknuntes lo katholou dia lou de/on einai lo kath'hekaston) -" It is exactly this initial project of scientific understan­ ding that is performed by the instances of preliminary explanation we have been considering, explanations which I have suggested are represented formally by the syllogisms of mixed modality of Prior A nalytics. It is as i f Aristotle envisions at the outset of all scientific understanding hybrid syllogisms of the form: triangle belongs contin­ gently to this figure; having two right angles belongs necessarily to triangle; therefore having two right angles belongs necessarily to this figure qua triangle. What this means is that in a limited but important sense we have in fact begun to explain why the angles of the figure before me are equal to two right angles, and why that light in the sky has gone out. We have done so just insofar as we have revealed the figure before me to be that sort of figure, or the light to be that sort of light and of that sort of body, for which we are then able to provide the true causal explanations by which necessity will be revealed: not until we have, but nevertheless

(21)

Poslerior Analytics ,

I

1,

7Ia8.

362

A. KOSMAN

insofar as we have. And this characteristically Greek " nevertheless" must be true if experience made intelligible is to be the ultimate source and telos of science. This last fact should finally help us realize why Aristotle introduces syllogisms of mixed modality with necessary conclusions. Consider for a moment Aristotle's parallel strategy in the Metaphysics. Like the platonists, he wishes to restrict substantial being to per se being and to restrict per se being to the robust formality of an entity being what it is. But unlike the platonist, he wants to resist the disseverment of that being from the entity which exhibits it, and to deny an ontology whose central image is that of entities separate from and therefore in relation to their formal being; at the heart of his ontology is the vision of substantial entities which are themselves. So here, Aristotle wants to restrict the world of understanding to the world of per se and necessary being. But he wants to deny that the world of the necessary is anything other than the world of the contingent understood properly in terms of its necessary rational structure; he wants, in other words, to resist a platonist separation of contingency from necessity. At the heart of all our reasoning about that intelligible structure of necessity by which the world is understood is the recognition of the incorporated unity of that structure with the dumb singularity of the particular.22 There is at the core of all explanation then, in the act the medievals called compositio, the synthesis of a particular with the necessary structure of its being, a recognition that some this is such and such. If we hope to account finally for this fact, we will have to introduce syllogisms of mixed modality. This is the insight that lies behind Prior A nalytics I 9, which introduces exactly such syllogisms and claims that their conclusions will be necessary, even though, as we learn in the Posterior A nalytics, they do not represent unqualified understanding. The subsequent tradition which joins Theophrasttls in denying validity to mixed Barbaras with assertoric minor and necessary conclusion courts the danger of the separation that Aristotle means to avoid; it risks precisely that dissociation of particularity and being that he saw as the chief difficulty of platonism. In one sense, this is all Aristotle's rabbit out of the hat trick, and mixed modal syllogisms with necessary conclusions, since they seem to pull apodictic rabbits out of assertoric hats, must seem to the Theophrastuses of the world like the most outrageous form of legerdemain. But it is in fact a rabbit that gets pulled out, and it was in fact in the hat. The magic in other words is genuine, that is to say,

(22) Small wonder that Aristotle was as attractive as he was to those religious traditions in which (one or another form of) hypostatic unity was mythically central.

NECESSITY ' AND EXPLANATION IN ARISTOTLE'S

ANAL YTICS

363

depends upon very skillful sleight of hand. The sleight consists in knowing how to understand the world properly, that is, in knowing what are the appropriate descriptions in terms of which individual entities come to be understood. The magic, that is to say, is the magic of nous, the sleight of mind by which we are able to come to recognize particular things for what they are. We began by suggesting that (x)[Sx --;. NecPx] represents the structure of some modal propositions. Perhaps, as we thought, that structure models the very mixture of particularity and necessity we've been looking at, as though the syllogism mirrored itself in the very structure of its proposition in ways we noted earlier. If so, the dual nature of modal propositions will seem not so egregiously an ad hoe hypothesis brought in to explain the difference in the validity of the two varieties of mixed modal syllogisms; the mixture of the modality of the syllogism will seem reflected in the mixture of the modality of the terms in the premises of the syllogism. But perhaps this is fanciful, and the truth is simply that Aristotle allows different structures in order to explain the different forms of reasoning which he wishes to come out right. Perhaps he is merely ambivalent. But if Aristotle is ambivalent in his reading of the logic of necessity, it is not because of any obvious muddle in his modal thinking. It signals rather his principled refusal to countenance any easy Academic separation between the world and its intelligibility, or between entities and their essential being, and his insistence upon grounding all explanation and all understanding in the intelligible world of particulars rightly revealed. This does not contradict Aristotle's claims that the particular cannot be understood. The reading I have suggested acknowledges Aristotle's insistence that they cannot be understood qua particular, but reminds us that in the act of understanding, the particular is apprehended into a general description in terms of which its features can be understood. Explanation, I have meant to claim, is directed ultimately to the particular, but to the particular revealed in terms of those general natures which explain its being, revealed most fundamen­ tally in terms of what it is. As a consequence, the hybrid modes of reasoning figured in the Prior Analytics do not in themselves generate scientific understanding, nor do they form any part of a finished discursive science. Like the perishable particulars they serve to explain, they do not last, nor are they meant to last, since they serve to reveal the world for us and vanish after science teaches us to recognize the world for what it is. On the other hand, the supposition that we might come to recognize the world for what it is an ideal vision of the regulative telos of science; in actual practice we remain situated in the world of the particular and

I :I" · '1

A. KOSMAN

364

perishable, and therefore always in need of the intercession of the Barbaras o f mixed nature. Unlike God, who dwells in the high and clean air of necessity, we continue to live in that peculiar mixture of necessity and contingency that is for Aristotle, as it was for the tradition he inherited, the distinctive climate of human being.23

A much earlier version of this essay was given at the University of Pittsburgh; I'm grateful for much helpful criticism then and duri? g di��u�sion at the 011'.lrons conference. I have been particularly helped by the searching cnbc1sm of James Lennox. (23)

T I I

B10LOGJE, LOOIQUE ET Mtl'APHYSIOUE "'"'

"""""' 1987

SCminaire CNRS-N.S.F.,

Editions du CNRS, Paris, 1990.

ETUDE PREPARATOIRE A UNE INTERPRETATION DU SENS ARISTOTELICIEN D' EIIArnrH* F. CAUJOLLE-ZASLAWSKY

La tradition est de rendre par « induction» le terme d'e1t1xywy� chez Aristote. Pourtant, cette traduction, qui, comme chacun sait, « a besoin d'etre accompagnee d'explicationsi>1, ne peut satisfaire personne, ne serait-ce que par la fagon dont elle privilegie indilment l'un des sens du terme - et, qui plus est, un sens qui n'entre peut-etre meme pas dans Jes diverses acceptions proprement aristoteliciennes. Cependant, la difficulte ne consiste pas tant a determiner ce qui separe 1'<1t1xywy� d'Aristote et !'induction moderne2, qu'a comprendre l'usage aristoteli-

(*) Une premiere version de cet article a ete presentee en juin 1987 aux Rencontres Internationales d'Oleron sur la Biologie d'Aristote, qui ont eu pour objet, cette annee-13, du 29 juin au 3 j uillet, I'etude des rapports entre methode scientifique et metaphysique aristoteliciennes. (N. B. : Jes chiffres entre crochets droits renvoient aux ouvrages dont on trouvera la liste dans Jes Indications bibliographiques, a la fin de cet article). (1) P. PELLEGRIN, La classification des animaux chez Aristote, Paris : Les Belles Lettres, 1982, pp. 1 4 1 , 150, 180-3. L'auteur rejette, notamment, l'idee regue d'une «contruction inductive des families animales» par Aristote en montrant que !'absence, dans sa methode, d'observation ou d'experimentation au sens scientifique du terme, fait qu'on ne saurait trouver chez Jui d'«induction1> proprement dite. La traduction d'bto:ywyf/ par « induction » est, d'ailleurs, critiquee de meme par la grande majorite des @mmentateurs. J. Hintikka, dans une formule qui semble resumer l'avis general, parle de « lranslalion highly misleading », [9), p. 423. On conserve pourtant cette traduction traditionnelle, faute de trouver mieux. En fait, ii ne s'agit pas reellement d'une traduction, mais d'une copie de copie : le latin a reproduit &7t-o:ywyfj dans in-ductio, apres quoi le frangais a fait de meme avec in-duclio, « induction ». La forme du mot est transmise, ma.is le sens n'en est pas plus clair : « conduite sur)), ou « vers», ou « contre1>, peut s'entendre de bien des fagons. (2) Selon T. Engberg-Pedersen [8], ce qui separerait principalement l'bto:ywyfj - de !'induction au sens oU est pris aujourd'hui ce terme, c'est la presence, clans le cas de !'induction, d'un ,critere de validite : !'induction moderne contiendrait l'idee d'une

I! · .' I ,[,,

·•.'1 1'1 I.·;• .' !i'.. \ l.' ' '"

li: 1cl 1 .•

li "'

\:

I i'

,

I

li 1

1\!

ii! 'rl

·1'l1_.

.. 1

''

F. CAUJOLLE-ZASLAWSKY

INTERPRETATION DU SENS ARISTOTELICIEN D'EnArO:rH

cien Iui-meme, qui se perd au travers d'une multiplicite de sens disparates - « convaincre par des exemples 1> et « inferer l� prin�ipe . general » constituant, de l'avis le plus repandu, la divergence semanllq� e essentielle -, pluralite qu'on meconnait d'emblee en rendant un1forme­ ment le terme par
second sens, E:7tiXye�\l au moyen signifie, d'une fa<;on genera1e, se procurer, pourvoir a ( . . . ) pour soi-meme, qu'il s'agisse de ressources, d'allies, de temoins, d'une aide ou d'un secours quelconques. On retrouve done, au moyen, certaines des acceptions que nous avions relevees pour la forme active du verbe6• Pour le substantif d'bto:yroy� maintenant, ii signifie globalement toute action d'apporier quelque chose, d'apporter ii, OU l'action de mener ou amener quelqu 'un a - un sens general qui se monnaie en une multiplicite de nuances particulieres : l'aclion de faire venir a son aide (on amene a soi) mais aussi l'invasion et l'allaque (on apporte la guerre, on mene une armee sur) ou alors l'ailirance, la seduction, l'allrait (une autre fa<;on de mener a soi) ou encore l'incantaiion, la formule d'envoU.tement, le sort, le malefice, le « charme » (une menee contre ou l'acte d'orienter sur) ou enfin la deportation d'un caplif en esclavage, en captivite (sans doute au sens d'une prise de guerre que le vainqueur «apporte » chez lui), d'oU, plus generalement, delresse et misere (sens derive, qui n'a plus grand­ chose a voir avec le sens initial) 7• On a done affaire a un usage courant d'une extreme richesse, d'oll ii ressort en particulier que le terme possede, par son indetermination meme, une capacite semantique hors du commun. C'est done le contexte

366

Notons d'abord que la diversite de ! 'usage aristotelicien ne fail que refleter celle de !'usage courant, dans lequel se rencontre deja, en effet, une profusion de nuances. Elles viennent differencier - la construction syntaxique et le contexte aidant - un sens general peu determine. C'est que le verbe E:7tliyEi\I a un sens general si large (mener Ct) qu 1 il p� ut, en effet, s'appliquer a des situations multiples et s'adapter aux domames les plus divers, en revetant des sens tres eloignes les uns des autres . On distingue ainsi 4 un premier sens1 oil €7tOCyeL\l signifie apporler quelque chose, l'apporter a que]qu'un par exempJe, OU bien a Un endroit determine, ou encore sur (contre) quelqu'un ou quelque chose, ou amener quelqu'un sur ou contre quelque chose, amener quelqu'un a quelque chose, appliquer quelque chose a quelqu'un, a quelque chose'. En un

consequence, c'est-U-dire d'une inference, du particulier U l'universel, ce qui entrainerait la necessite d'une validation du procede inductif; au contraire, ii n'y aurait aucun mouvement d'inference dans l'bta:ywy�, meme dans Jes Analytiques premiers, II 23. (3) Engberg-Pedersen [8] distingue six sens d'£:1ta:yw0/, qu'il tente de reunifier. I-Iintikka [9] propose, quant a It.i i, de reunifier en tout cas Jes sens oU le terme designe une maniere de connaitre les premisses d'une science. (4} Nous avons suivi, en gros, LIDDEJ.L/ScoTT/JoNES, A Greek-English Lexicon (reimp. 1973), mais, d'une fagon generale, les usages poetiques n'ont pas ete retenus, cependant que de nouve!)ux exemples etaient ajoutes, utiles a notre recherche. Pour ces complements, on a utilise les ouvrages suivants : H. Bo N TTZ , Index Aristotelicus, Gratz : Akad. Verlag., 19552 (Berlin, 1870) ; J . E. PowE LL, A Lexicon to Herodotus, Hildesheim : G. Olms, 1977 (reprod. Cambridge 1938), p. 126, art. krt&.yw, Ertocywy6� ; E. DES PLACES, Lexique de la langue philosophique et religieuse de Platon, t. XIV de Platon, CEuvres completes, Paris : Jes Belles Lettres, reimp. 1970 (1964), 2 vols ; K. JANACEK, Indices; vol. IV des Sexti Empirici Opera, Leipzig : Teubner, 1970 ; A. BF.LIS, Aristoxene de Tarente el Aristote : le traile d'harmonique, Paris : Klincksieck (<( Etudes et commentaires» : 100}, 1986, pp. 212-215, kTCocywy�. Le texte d'Aristoxene est celui de M. MEIBON ed. Antiquae Musicae Auclores Septem, vol. I, Elemenla, pp. 1-74, Amsterdam, 1652 (avec traduction latine et notes}. (5) TOv '{TCrtOV e. : Herodote, c. Hude ed., Oxford (OCT} : 3, 85; x�vS:Uvou� ·nv[ : Isocrate, F. Blass ed., Leipzig : Teubner, 1889-98, 8. 3 ; amener des chiens sur la trace du gibier : XENOPHON, Cynegetique, 10.19 : -moy. ib., 6. 25 ; conduire une armee contre l'ennemi : ·nv&: Ertl 't'�va:, Thucydide, H. Stuart Jones, Oxford (OCT) : 8.46 ; d'oU aitaquer (absolument, Ht'lrodote 1 , 63 ; 7, 157, 165 ; 8, 1 12) ; presser le mouvement, acceterer : PLATON, Cralyle,

367

420 D ; e&:t"'t'OVct. Pu0µ0v e., XENOPHON, Banquet, 2. 22 ; mener quelqu'un par son influence : Thucydide, 1 . 107 ; avec l'inf., inciter quelqu'un a faire quelque chose : Isocrate, 14, 63 ; invoquer !'aide ou !.'alliance de quelqu'un, faire enlrer quelqu'un dans une alliance, le prendre pour a!lie ('t'Ov IlEpcniv, Herodote 9, I ; T0v IlEpcrYJV ETCl Toll� "EAAY)va:�, DEMOSTHENE, Lettre de Philippe, 12. 7}; faire uenir, procurer (hn-riiS:e:toc, Thucydide 7. 60 ; TiX Ex TWv S:LwpUx.wv e. v&µa:-ra:, PLATON, Critias, 1 18 E ; metaphoriquement, krt&.ye:� � 1iux� -rO f:v lf).J.,, Plotin 6.9. 1 ) ; etendre OU allonger sur, appliquer a (nA71nv knl 't'tVa: : La Septante, Isai'e, 10. 24 ; £. �71µla:v kTC�'t'L0&vct.L : LUCIEN, Anacharsis, 1 1 ; e. tjv S:t.Xvot.Xv 't'tvt, PLUTARQUE, Pericles, 1 ) ; u n usage comparable a !'usage platonicien d u verbe krtt8oct�, Lois, XI 933 D 7 ; 'Ex&'"}� o:p&crx.wv kTCa:yw�v yr:yov£voc�, TttEOPH �ASTE, Caracteres, 16. 7. =

368

F. CAUJOLLE-ZASLAWSKY

qui, en chaque occurrence, va etre decisif pour determiner le sens precis. _ Ainsi, le terme signifie l'attrait, la seduction, clans le vocabula1re amoureux, l'absorption de nourriture dans le vocabulaire biologique, !'insertion de jours intercalaires dans le vocabulaire chronologique, !'induction au sens medical (!'induction au sommeil par exemple) dans le vocabulaire de la pharmacologie, l'appel a temoin ou, suivant le cas, !'introduction d'un temoignage, dans le vocabulaire juridique, la presentation d'un nouveau membre a une assemblee dans -le rituel politique, !'incantation dans le vocabulaire de l 'occultisme, !'importa­ tion dans le vocabulaire commercial. Mais ii faut m€:me1 parfo1s, au se1n d'un seul domaine, restreindre encore le champ et determiner des sous­ contextes, si l'on veut saisir adequatement l'acception du terme. Car, dans le vocabulaire militaire, par exemple, ii signifie l'invasion, tout en designant aussi, d'un point de vue tactique, une formation particuliere d 'attaque8, et il signifie en outre, toujours dans le contexte guerr1er, . mais dans la perspective de l'intendance, l 'apport de subsides ou de vivres, et dans une perspective strategique, l'acquisition d'un allie ; enfin, nous l'avons vu, ii signifie, egalement, la deportation des prisonniers. Le terme nous promene done, en gros, de !'invocation d'un temoignage a }'invocation malefique, de !'introduction a !'invasion en passant par !'intrusion, de !'importation a la deportation et de !'induction au sommeil a !'induction logique a travers }'induction a l'erreur et !'induction a mal faire. _On voit la place considerable qu'il . faudra faire, en consequence - et meme si nous pouvons d'emblee ecarter un certain nombre des acceptions ci-dessus pour l'E7tixyc.uyfi qui nous occupe ici -, au travail d 'interpretation. Le contexte, en effet, ne

Ii I

I

INTERPRETATION DU SENS ARISTOTELICIEN

D ' EilArOrH

369

pourra, eventuellement, trancher du sens de fagon decisive que s'il est d'abord etabli lui-meme avec exactitude. Et c'est precisement ce point qui, selon nous, constitue la premiere difficulte pour comprendre le sens d't7t&:yew et d'&mxycuy� chez Aristote. Car il ne suffit pas de determiner si ces mots prennent sens dans un contexte logique, epistemologique ou pedagogique : il faut tenir compte de leur possible evolution semantique, afin de ne pas projeter dans les ecrits aristoteliciens une acception plus tardive. Il est done necessaire de faire intervenir aussi !' aspect historique du contexte. Dans cette derniere perspective, il nous faut signaler un temoignage de Sextus Empiricus concernant l'&mxycuy�. L' interet de son texte est de nous montrer que le privilege accorde au sens de
(8) Dans la langue militaire, l'bi:cxywy� fait reference a une formation tactique particuliere, dans \aquelle une aile est disposee a la suite de l'autre par contraste avec la formation appelee n«pcxywy� (AscLEPIODOTE [le tacticien], Aeneas tacticus, Asclepiodotus, Onosander, W. A. Oldfather ed., Collection Loeb, 1923, 10. 1 , 1 1 .24). D'autre part, l'kn«ywyfj designe, dans le domaine biologique [en dependance du sens du verbe : apporter quelque chose a un endroit particulier (&µcx�ocL .. 't'oU<; AWou<; btljyov, Thucydide, l . 93) qui, au moyen, s'applique, par exemple, aux racines qui font entrer, tirent la nourriture a l'inlerieur d'une plante (THEOPHRASTE, Histoire des planles, 1 . 1 .9), et se retrouve, analogue, au passif ('t'pocpOC Sn&:ye-ri:u 't'(i) aWµoc·n, Timee de Locres, C. F. llermann ed., Leipzig : Teubner, 1 852 : 102b) !'absorption de nourriture par le gosier (ainsi chez A RISTOTE : de Spiritu, 483a9), mais aussi la direction dans laquelle se fait une pousse, celle des cheveux, par exemp\e, Yi -ri'j<; -rp�z6<; Snocywyfj (Diodore de Sicile, 3. 35) ; en medecine, le verbe kniXye�v signifie induire au sens d'entrainer, de provoquer : bi:&:yeiv -rljv xoiA(ocv (D1oscoRIDE, De Materia Medica, 4. 157, M. Wellmann ed., Berlin, 1906-1914, uaria lectio pour Un&:yew). Cf., chez Platon, l'adjectif bi:ocywy6<;, au sens medical de <(adducteur)> ou <(inducteur» btocywyOv Umou, Timee 45 D 7. Enfin, en liaison avec le sens apporler en supplement, en sus d'kniXye�v, on a le sens technique particulier intercaler des jours (n£v-re Yiµ&pcx<;, Herodote 2, 4 ; ext €nocy6µevoci, .avec ou sans i-/µ&ptXi, Jes ours intercalaires, D1onoRE DE S1CILE, I . Bekker et al. eds., Leipzig : Teubner, 1888-1906 13).

(9) SEXTUS EMPifUCUS emploie, d'une part, la forme €n1Xywyix6r;, inducfif (generalisanl) : Hypotyposes pyrrhoniennes, II 196 (0 e. -rp6no<;), 213 (Snocywy�x� €nto-TIJµ1J) ; Cantre les professeurs, IX 95 (8Uvocµ�<; €nocywyLx1}) ; et kncxywyixW<; (au sujet des propositions etablies de maniere inductive) HP II 195 197 (knocywytxW<; �eOcxtW). II emploie egalement le =

. . '

I .:

i

I

I,'

370

I

INTERPRETATION D U SENS ARISTOTELICIEN D'EIIArnrH

'

F. CAUJOLLE-ZASLAWSKY

Curieusement, cette image de l'btocywy�, dont on peut dire d'embl ee ne - meme en ignorant le sens exact du terme chez Aristote - qu'elle cus Empiri Sextus urs, {d'aille elicien saurait refleter aucun usage aristot ne commet pas l'erreur d'attribuer au Stagirite l 'absurdite qu'il decrit) dit est devenue ensuite dans !'opinio n le type meme du raisonnement de texte le et ele), pele-m on», inducti « et sme « aristotelicien » (syllogi aussi eSl>, Critiqu « es fameus des e l'origin a meme de Sextus se trouve incongrues qu'obstinees, de la pretendue circularite du raisonnement aristote licien. ­ Comment pretendre, cependant, qu'Aristote a voulu « faire inducti pede' quadru vement I ' assertion de la premisse ' aucun homme n'est un , une en partant des particulieres )) ? Dans une perspective aristotelicienne e, Aristot pour n, questio telle chose n'a pas le sens commu n. II n'est pas et eux, d1entre e nombr grand de passer tous les cas en revue, ni meme un a - avant toute autre raison - parce que cela impliquerait de traiter d'un pieds Jes r compte t faudrai egalite l'accidentel et l'essentiel : ii non animal sans consideration du fait que son nombre de pieds est dil ou pied, un a es d'homm e nombr : s colonne a un acciden t. On etablirait des Or, nombre d'homm es a deux pieds, nombre d'hommes a trois pieds, etc. que ation signific meme la pas n'a cela , homme un avoir deux pieds, pour on, d'en avoir un ou pas du tout (dans les deux cas, ii s'agit d'une privati ent equival un comme compte etre peut ne on privati d'une et le resultat de l'etat normal), pas plus que d'en avoir trois ou quatre. Non point de qu'etre bipede soit un caractere essentiel, qui entre dans la definition avec ne etre nt l'homm e ; car, si un homme se trOuve accidentelleme Jes quatre pieds, cela n'en fait nullem ent un quadrupede (sauf a jouer sur homme aucun « ition mots), pas plus que cela ne rend fausse la propos n'est un quadru pede » : pour Aristote, un homme qui aurait quatre pieds avoir serait simplem ent un homme atteint d'une malformation. Brei, tote qu'Aris ce de partie e, l'homm deux pieds ferait, dans le cas de de voie par « mais soi, de vertu en appelle Jes proprietes qu'on possede triangle le pour droits deux a egal d'etre consequence»10 - comme c'est une propriete par soi, mais deduite (et non pas « indui,te »). �

Le texte de Sextus Empiricus, tout en nous indiquant I' evolution du t�rme, fa1t apparaitre que l'induction ainsi comprise ne saurait etre celle . d Anstote. On considere generaie rr:ent que, dans la logique d' Aristote, £7t&:yetv . . s�gn �_ fie e?setgner ou convazncre par induction ; mais le ve· rbe semble s1?'n1fier egalement : meilre en avant, avancer, d'oiI inferer le principe , g� neral". Quant a l'bcaywy�, on l' interprete done, dans Jes textes d Ar1stote, au sens de raisonnement induciif, « induction » en se fondant . . sur la descr1pt1on aristotelicienne la plus connue de l'€7tixy� y+,., ·. £7tixywy1) 1) ' , " IX7tO TWV xix6' exix �Tov e7tl Ta xix66Aou E�o8oc:;12, qui est, en effet, la seule sur laquelle on pmsse se fonder pour tenter de defendre Ja traduction _ « Induction». L'€7tixywy� serait done un passage ' une voie d'acce's (e�oooc:; " • ), ' . . . qu1 menera�t, d epu1s l � s � as 1nd1v1duels, aux cas universels. Notons que ,, ce sens, ou l <7tocywn s oppose au syllogisme en tant qu'il est un _ ra1sonnement deductif (d'oU, par contraste, la traduction d'« induction » �our « k�ixy�y� ») est le se �s q� i, par la suite, prevaudra, comme l'indique l emplo1, c�-dessu � ment1onne, du terme chez Sextus Empiricus. Ge qm nous mteresse, ic i, est d 'observer de quel passage ii s'agit _ exactement et en qum 11 cons1ste, C'est qu'Aristote dit ailleurs ' en effet e la meme lmocywy� qu'elle est le principe, le point de depart {ocpx�) d � l un1versel13. �t nous avons 13. une premiere divergence entre deux sens dans la relation de l'€7tocywy� a l 'universel : un « passage » est u � processus, et ce ? 'est done pas � ne base de depart, mais ce qui a lieu _ entre pomt de depart et pomt d amvee. L'e"aywy� est-elle ce dont on part pour atteindre l'universel ou bien l'une des voies qu'on emprunte pour acceder a l'universel ? Essayons de suivre, tour a tour, les deux directions et voyons oU cha ?une d'elles va nous mener. Commen�ons par explorer la voie oU la not10n d'e""'Y"'Y� se trouve Jiee a celle de principe. Or, sur cette piste, '

•

•

·

·

·

�

(I 1) Presenteraient le premier sens les passages : Top;ques ' V I I I l 1"6 .., a4 .' au ' ' . · A l t ° S pass1r na ¥ 1ques econds, I, 18, 8lh5; cf. I, 1 , 71 a21 , 24 ; absolument : Rhetorique 1356b8 ; Toptques, V I I I , 2, � 57al8-21 sqq ; 14, 1 64a l2-13; Analytiques Seconds JI, 5, 91h15 '. Le second sens se trouvera1t, par exemple' en Top;ques I 1 8 l08b l l , etc. De 1a· para1't · I e sens que ce verbe prendra par la suite : ajouter une raison, conclure, infkrer. ven1r . (12} To� iques, I, 12, 105a l3. Pour !'ensemble des textes d'Aristote relatifs a ,I €1t�'YWY1J, vo1r H. BONITZ, Index Aristotelicus, pp. 263-4). Voir egalement w D Ross (Ar1slolle, Landres, 1 923, 1 9496, �p . 38-41 ; et !'edition des Analyliques, Oxfo�d " 1949, pp. 47-5 � et 481-483) et Jes notes a1ns1. que I , introduction, de J. BRUNSCHWIG 3. son edition des Topiques I-IV, [Arisfofe, Topiques, T. I, L. I-IV, Paris : Jes Belles Lettres 1 967 . �otam ?1ent pour les passages : Topiques I, 8, 103bl-l9 ; 12, 105al0-20 « Dedudtion ei indu ct10� » ; � I , 8, 1 � 3b � 7, 29 (&.v&.7to:Aw) ; 10, 1 1 5a5-6 (si l'accroissement du sujet entraine ?elui. d � I accident, I accident appartient au sujet) ; IV, 2, 122a1 9 (autre cas montrahle par ·

•

substantif : HP I I xv 204 : 7tspl E7ta:yw�i;), 't"Ov 7te:pl k7tixywyjjt; Tp67tov, procede dont ii decrit !'intention de la sorte : « &:7t0 'TWV wx'Ti!t µ£pot; 7ttG'C"oii0"0a:t "oUAov't"i:n 8t' a:Utjt; TO xix06Aoui>. Voir aussi : Ibid., eh. xvii (7te:pl 8tixtpE:O"swt;) 213 : (j 'E7td 8£ 'Ttvi::t; 'TWv 8oyµct'Ttx.Wv TI)v 8ta:Ai::x'Ttx1jv i::lvix( 195-197 (qu'il tpctO"tv k7tta't'f/µ"l)v 0'1JAAoytO"Ttx1jv btixywytx1}v 6pta'Ttx1jv 8tixtpi::'Ttx.�v . .. ll, et /IP I I serait instructif de comparer avec Analyliques premiersEI I 22-23). De 13., ensuite, le sens : ed., Leipzig : auAAoytaµoUt; � E7ta:ywyO:t; 7ti::pix(vov'TIXt;, chez Polystrate l' picurien (C. Wilke etc.). De 13., 2.4.6 exemp\e, (par Plotin chez aussi trouve Teubner, 1905, p. 1 1), sens qu'on egalement, le sens de raisonnemenl dialectique, entendu comme etant destine a conduire l'adversaire dans un piege (chez Aulu-Gelle, par exemple, 6 (7).3.35 ; ou Diogene Laerce, I I I 53). (10) Nous faisons allusion a la notion de O"Uµ6e:6"1)x.6Tix x.a:6' a:U'T&.. Cf . Meiaphysique B, 997a7 : I'>, 30, 1025a30-35 ; Z, 4, 1029bl6-23.

371

•

•

3, l23b7-8 ; et pp. xxxn, n. 2 ; 125 (p. 12, n. l ) ; 137 (p. 32, n. l ) ; 149-150 ; 153 _ : 100a25 et notes, pp. 1 1 3 ; 125 (p. 12, n. 2)]. (p. 57, n. 2) _, �our la deduction (13) Ethique a Nicomaque, VI 3 l 1 39h28, par exemple. rn•rw"Y'I) ;

I

.'!

373

CAUJOLLE-ZASLAWSKY

INTERPRETATION DU SENS ARISTOTELICIEN D'EIIArnrH

survient une indication qui ne clarifie pas les chases, celle que, parmi les principes, certains sont pris en consideration au moyen d'une E7tocyooy-fi (btocywyjj), d'autres par une sensation (oclcrO-ljcreL)14• Nous voila done avec une E7tocywyfi qui n'est plus un principe, ni d'ailleurs un processus, mais un acte simple de voir Jes principes (Oewpei:v). Et de plus, cette « vision » est presentee comme n'etant pas d'ordre sensible ; elle s'apparenterait alors a une sorte d'intuition intellectuelle 1 5. Principe OU intuition des principes ? Nous voila dans !'impasse. . Mais, ce premier essai, pour etre infructueux, nous a cependant permis d'observer la repetition d'un certain schema : en deux occurren­ ces successives, l'E7tocyc.uy-fi est apparue comme se rapportant de deux fagons incompatibles a une notion centrale. Dans le premier cas, la notion est celle d'universel : l'E7taywy-fi a trait a l'universel, mais tantOt comme son principe et tantOt comme sa voie d'acces. Dans la seconde hypothese, l'btocywy-lj a trait a la notion de principe, mais tantot elle est un principe elle-meme, et tantot elle est !'intuition des principes. Essayons de reduire autant que possible ces contradictions. Un moyen de le faire serait eventuellement de comprendre ainsi Jes chases : l'btocywy-lj serait principe de !'universe! au sens de point de depart, c'est­ a-dire d'oppose. Ce serait, dans cette perspective, la sorte de particulari­ te qui donnerait acces a ]'universe! (car toute forme de particularite n'est pas apte a le faire) . Et peut-etre la nature particuliere du point de depart impose-t-elle aussi une voie particuliere d'acces a l'universel : le point commun a l'Enaywy-fi comme base de depart et processus serait Sa particularite, laquelle constituerait, en !'occurrence, le trait le plus caracteristique de l'btocywy-lj. II reste a trancher pour elle entre la nature d'origine ou de processus. Appliquons le meme traitement a la seconde contradiction : « principe de l'universel » ou « intuition des principes » ? L'oppositioh se reduirait si !'on substituait a la designation trop vague de « principe» ses deux acceptions precises, celle de commencement et celle de commande­ ment : l'btocywy-lj serait point de depart vers !'universe! comme intuition d'une regle. L'indication du caractere de particularite se �etrouve ici dans la mise en parallele avec la saisie par la sensation - qui est l'autre fagon de saisir ce qui est d'ordre particulier. Notre hypothese est done que l'hcocywy-lj est la vision, non point sensible mais intellectuelle", !'intuition intellectuelle, sur la base d'elements particuliers (rien n'empeche qu'ils soient eux-memes sensi­ bles), de quelque chose qui n'est plus particulier. Mais ii faut, pour

�roc_e�sus, do�ne� acces a l'universel. Ou alors, ii faut imaginer que cette

372

F.

(14) Ethique a Nicomaque, I, 7, 1098b3. (15) Voir, sur l'
evacuer toute contradiction, que cette intuition puisse, sans etre un

mtmt10n, qm fa1t acceder a !'universe!, si elle est un point de depart relativement a ce dernier, est en meme temps situee au terme d'un proc�ssu � . cette intuition ne serait pas immediate, mais provoquee, . �onsecut1v� a une recherche, a un apprentissage. Ce serait, par exemple, a un certain stade de la reflexion, la decouverte, !'intuition au sens de vision intellectuelle (comprehension claire, restructuration soudaine) d'une forme conceptuelle evidente, OU d'une regle. Aristote pourrait avoir designe du meme terme la demarche methodique procedant par Jes cas particuhers concrets et le couronnement intuitif de ce type d'approche. L'btocywy-lj serait done, dans cette hypothese, a la fois un processus menant a un acte instantane d'intuition (la perception mentale d'un umversel, par exemple d'une regle applicable a tout cas particulier de meme sorte) , �n procede (la methode qui s'appuie sur le comportement des cas particuhers concrets) , un point de depart vers !'universe] (different de l'etablissement des premisses du syllogisme) . On voit qu'il est done possible, dans ces conditions, d'interpreter l'btocywy-lj dans differents contextes : dans une perspective pedagogique elle peut consister a faire passer un eleve d'un etat d'ignorance a u � �a:oi�, ou _bien a passe: . r�tionnellement et dans une perspective ep1stemolog1que, de la vahd1te de propositions particulieres a !'assertion d'une proposition universelle17•

(17) �� pre�iere inter�re�ation est celle, notamment, de K. von Fritz [2], qui . rappro?he I �na.yw"fl'l d� la � a1eub �ue te_lle que la pratique Socrate aupres du petit esclave ?� Men�n . « e:na.yw"fl'l, c _es�-8-d1re (!Introduction» (Heranfii.hrung) au sens le plus elementa1re de ce mot; ma1s Il y a aussi des cas de conduite de la preuve (Beweisfii.hrung) dans le.squels entrent Jes elements de la deduction au sens moderne. Cela est _ , part1cuhereme nt clair da�s Jes cas mis en avant par PLATON, dans le Menon (828 sq.), oil _ 1 o rate ? , condu1t l escla:e _1ncult.e - et c�la en l'i�troduisant (heranfii.hrl) au probleme avec �I aide d une figur� dess1nee - a parven1r par lu1-mllme, apres plusieurs essais infructueux, a admettre que, s1 I ,on veut doubler la surface d'un carre, de fa4$0n que Ja surface doub!ee _ a. nouveau la forme d 'un carre, on doit obtenir le carre construit sur la diagonale a1t (p. 20� )- ARISTOTE pa�le �a (en R�et�riqu�, 1 , II, eh. 20, I393a32sq.) de !'usage du P?rad1�me da �� la rheto�1q�e, . et , 11 1de_nbfie cet usage a celui de J'epag6ge dans la d1alec;1que. lei egalement ii s �git d une «introduction » (ein Hinfii.hren) des auditeurs a ce q� e. I orateur veut leur suggerer (p. 217). Mais deja Jes deux termes techniques qui des1g�ent l�s. deux metho?es de s'�ssurer la certitude Jes plus originelles, epagOge et . des1gnent un�
\I·

: i'' J

374

Et, s'il s'agit d'un precede de ce genre, Ja question se pose du nombre de cas necessaires a son bon fonctionnement : combien d'exemples faut-il pour convaincre ou pour generaliser? Combien de cas particuliers sont-ils requis ? Et selon quel critere certains cas particuliers seront-ils choisis plutot que d'autres18 ? Enfin - et nous nous en tiendrons la dans !'enumeration des difficultes --, quel est le rapport exact de !'btocywy'ij au syllogisme ? Comment peut-elle s'opposer au syllogisme si, « d'une certaine maniere >), elle est un syllogisme elle­ meme 19?

,,

:i I

'!

Nous nous proposons done de revenir une fois encore sur cette notion evasive, mais en empruntant pour le faire une voie un peu detournee : celle d'une enquete qui, avant d'en venir a l'emploi du terme dans l'ceuvre d'Aristote, porterait d'abord, d'une part sur son usage courant, d 'autre part sur certains de ses usages techniques ou savants, chez d'autres auteurs qu'Aristote. Car ce dernier n'a pas cree le terme d '£7tcx:ywy�, s'il est vrai qu'il en a marque l'usage en se l 'appropriant : il a trouve le mot soit, directement1 dans la langue ordinaire, soit l'a emprunte a un langage deja specialise ; et ce terme a encore ete utilise apres Jui. L'intention du detour que je propose est de verifier si d'eventuels usages d'btocywy'ij, independants ou differents de celm d 'Aristote, ne pourraient nous rendre ce dernier plus clair. Revenons un instant sur les acceptions courantes du mot E:7tocy<0y� pour constater1 en premier lieu, que la langue courante, contrairement a !'expression philosophique, ne met pas ! ' accent sur la notion et ne privilegie done pas le substantif btocywy'ij, mais donne Ja preseance au verbe. Le sens global du compose E:7t-6:ye:�\I est, litteralement, nous l 'avons vu, « mener sur», et !'ensemble de ses acceptions particulieres dans le Jangage courant est forme des interpretations variees auxquelles donne lieu la combinaison de ce verbe avec ce preverbe. Or, on sait que le preverbe E:7tl « n'esl jamais susceptible de se vider de son sens pr'·opre. Exclusivement « plein », il conserve a peu pres toutes les valeurs qui ont ete indiquees (§ 522 sqq.) Jorsqu'il a ete etudie comme preposition : ii n'y a guere de difference entre &7tl �oclvw Y�• et �oc(vw t7tl Y�• »20 , Le sens du

ii'

I 11· : i li i . ·I

I

t

'II '

I

I! I , 1' 1I .' ;

i

l'1

J

'I'' . II ' _ .,': ,_J,

:.f\· 11 1

:

, II_ _

<_,i:1.11 ·I

·1 1 1 : 1. · ·1' .1.' '

·

�-.', II :I ,.. '".'\·I· ', ·,. . 1f.·l' 1.· · 1· [ :rj 1�·I, .I1·'1 ._·!,11 1 ·: : '" I ' j [i!!

'.

:

:� !,

'' '

.

F , CAUJOLLE-ZASLAWSKY

•.

i

(18) Ce qui resoudrait la question de savoir s'il convient de separer des autres textes relatifs a l'&ncxywyfj ceux des Analyliques une telle separation n'est pas utile. (19) L'&ncxywyfj s'oppose au syllogisme en ce qu'elle n'est pas une demonstration : µcx.v6&voµe:v 1J Encx.ywyj'j 1J &:no8e:l�e:�, Analyliques seconds 81 a40 ; « 1h8cxaxcxAtct. ... 1j µEv 8�' !ncx.ywy�c; 1i 8E au/.Aoy�aµijl», Ethique a Nicomaque, 1 1 39b27 ; �O"'t'L 't'o µEv ncxp&:8e:Lyµct. &ncxywy-fi, :o 8' &v66µ11µo: auMoyLaµ6c;, Rhetorique, 1356b3. Toutefois, elle ne s'oppose pas au syllog1sme comme !'induction a la deduction : ii y a , comme le signale avec raison von Fritz [2], un aspect incontestablement deductif dans l ' Encx.ywy-fi, cf. ci-dessus, n. 17 : «ii y a aussi des cas de conduite de la preuve (Beweisfiihrung), dans lesquels entrent Jes elements de la deduction au sens moderne. » (p. 201}. (20) J. HUMBERT, Synlaxe grecque, Paris : Klincksieck, 19728, p. 337, § 598 (le soulignement est de l'auteur lui-m�me). Les valeurs de En( : au sens concret, btl avec

INTERPRETATION DU SENS ARISTOTELICIEN D'EilArnrH

375

verbe est des !ors interessant pour eclairer celui du nom (a la difference de &.\la.Aoyl�e:cr6oc�, par exe rr:ple) et ce sens sera en outre indique, quand le verbe ne sera pas em� lo�e abso u �ent, par le « cas.)> de ses complements. Quant au substant1! '""'Y"'n, 11 presente une diversite de sens c�mparabl e a celle _du verbe, a cette difference pres - qui _contribue d a11leurs ?_ � b � curc1r encore I� sens philosophique du terme - que Ja valeur e '".' n est plus md1quee comme elle l'etait pour le verbe, par le « cas» d un eventuel complement. II faut desormais se contenter de Ja simple indication suivante : « . .. E:7tl exprime avant tout l'image d'un contact avec une surface »21 _ Ce point ayant ete precise, venons-en maintenant a retude d'un usage plus technique du terme.

�

�

II a existe, en effet, avant Aristote un usage qu 'on pourrait qualifier de « pedagogique » des deux termes'. On le trouve, notamment1 chez Platon . Da �s un enseign em �nt ou une argumentation, E:7t6:yi:: �\I signifie, de _ a par exemple, chez Platon, amener des eieves , conduire fac;;o n generale, a percevoir ce qu'ils ne savent pas encore discerner, ou, chez Aristoxene de Tarente, amener les gens a percevoir, en le leur faisant ecouter 1 la be � u �e d'un � cc�rd dont ils c ?ntestaient, a priori et sur un p lan " harmomeux. Dans cette perspective, !'bt0tywy1J theorique, q � . 11. put etre est Un procede pedagog1que Consistant ii instruire OU ii COnVaincre quelqu'un en le confrontant aux evidences de la perception sensible". Notons, po � r memoire, que Platon emploie le terme d'E:7tocywy� aussi . b1en clans certaines de ses acceptions courantes que dans un usage plus

:

l'accusatif signifle «vers » ; E�l avec le genitif signifie «sur)>; Enl avec le datif signifle «II cote de, contre f>. Valeurs abslrades, au genitif (partitif) est Jiee l'idee d'autorite exercee s quelqu'un ; a� datif Jocatif d � proximite sont liees Jes idees de survivance, d'adjonction, �: cause, d� datif proprement d1t releve le fait qu'une action est accomplie quelqu'un («dativils commodi et incommodi»). , designant un raisonnement qui vient apres un autre ou en addition a s1gn1 u ? autre (dans Ausgewtihlte Schriflen, Berlin : W. De Gruyter, 1968, pp. 285-290' reprod. d un compte rendu de 1957). (22) , PLATON ·nvcxc; Enl -rO: µ�nw ��yvwax6µe:va, Polilique, 278 A (nous reviendrons plus . sur I analyse de ce passage capital) ; ARISTOXENE OE TARENTE emploie \a forme loin _ du verbe : i<Enu.x6tv-rwv cxU-rWv>> Traite d'harmonique, p · 23, 2 e'd . Me'b 1 om vo1r c1passive dessus, note 4) ; « Enax6e: atv cxU-roTc;i>, ' Ibid. 23, 9. II emploie egalement le substantif corresponda�t : (j 8t&: rljv btaywy1jv 8€ tjv En 't'oti-ro ytyvoµE:v"t)v xa't'&: 't'Ov x.wptaµOv 't'Ov &.nO TWv &X>:wv _&:vo:yxat6v nwc; xal -rWv &AAwv Enarpiia8cxt cp ae:cuv f>, Traite d'harmonique, preambule des Prtnctpes, P· 4, (!. 14-17}; , Ibid., Elements, p. 53 (l. 21). Voir . de A : BELIS, op. cil., pp. 213-215. On constate de fagon particulierement c��mentaire exemples que la traduction i< raisonnement par inductioni> exige claire sur ces dern1ers toute une reinterpretation.

PP

:

T

t

U

yYj

(··

376

' ·. ' [

\1',

I \'.

I ',

i 1: '

i

F.

CAUJOLLE-ZASLAWSKY

INTERPRE'FATION DU SENS ARISTOTELICIEN D' EnArO:rH

technique et proprement philosophique. II utilise en effet le terme, au pluriel et en association avec E7tcp81}23 ou xoc:&.8ecrtc; ou e1_1core �oc't'&.8eaµ.oc;, dans l'acception courante de mag1que1 pour dire qu un charme est « amene sur» une personne et pese sur elle, la plagant « sous ! 'influence » de quelque puissance occulte. Les exemples empruntes a Platon permettent dans une certaine mesure de preciser le sens du btl contenu dans &:7tocycuyfi : le terme est employe en association, notamment, avec E7tc.u0-fi, qui a le meme prefixe (x.oc-rocOE:cre:crLv 1) S7taycuycii:c; 1) 't'tcrL\I &7tcp3�Lc;). Or, la � aleur du E7tl de E7tcp8-fi se trouve etre clarifiee, danS certa1nes expressions, par le renforcement de ce preverbe au moyen de la preposition elle-meme : E:7tcp8-fi ae 't'Lc; S7tl 't'c'jl cpocpµOCxcp24• Nous rencontrons egalement chez Platon E:7tocycuy6c;, au sens usuel de « seducteur», et au sens technique medical de « adducteur)) OU (( inducteur)) : �7tocywxOv i'57tv�u 25 • . M aintenant, Platon emploie en un sens pedagogique bien precis la formule : &7t&.yEt\I "t't\ltx.i:; &7tl "t'0C µ�7tw yty\lwcrx6µE\ltx.. Dans le passage du Polilique ou nous la trouvons (en 278 A), ii est question de l 'apprentissa­ ge de I'ecriture. Platon part du stade ou l'enfant a une percept'.on suffisamment distincte (tx<Xvfu' 3'"''"6&:v."6"'' ; cf. Phedre 250 B ; Sophisle 253 B) de chacune des lettres dans Jes assemblages Jes plus courts et Jes plus faciles, et ou ii se montre capable de I es y d �igner correcte n;ent, , mais ou ii reste encore hesitant quand ii s agit d identifier ces memes Iettres dans Jes assemblages plus complexes, et commet alors des erreurs d'appreciation aussi bien que d'expression. Or, dit Platon, le moy;n le plus simple et le meilleur d 'amener (bt&:y«v) l 'enfant aux chases qu ii ne . sait pas encore discerner (E7tl "t'OC µ�7tw ytyvwcrx6µE\ltx.) cons1ste a le ramener, pour commencer, aux cas simples, oU il a identifie correctement les

�

(23) P. BOYANCE, Le culle des muses chez Les philosophes grecs, Paris, 1937 (pp. ,36-42) ; W. THEILER, dans Feslschrifl . . . K. Meuli. Schw. Archiv fiir Volksk., XLVII, 1951, P.P· 1967. (24) Charm;de 155 E 5, E 7 ; 156 A 2, 8 1 , D 4 ; 1 57 A 4, 8 4, C 4 ; 1 58 8 8 ; 1 75 E 2 ; J 76A 1 ; Euthydeme 289 E 5 ; 290 A 1 . Notons que !'usage de &1t48e:Lv semble indiquer u n sens assez voisin : £.1tetctTkov 1tp0 Tij.; -.oi3 cpct.pµ&:xou 86ae:w.;, Charmide, 158 C 2 (cf. 155 E 6, 176 B 2) ; XP� ..ix. '!'0Lct.i3'!'ct. &a1te:p £.1t48e:tV Ect.u"T�, Phedon, 1 1 4 D 7 (cf. 77 E 8 ; Republique, x 60� A 3 ; Theetete, 1 49 D 1 , 157 C 9 ; -Lois, I I 665 C 4 , 666 C 6) ; bt48ovrct. ... 1td0e:Lv 1te:tp : Tim€e 45 D 7. Les termes associes ou apparents aux precedents, chez Platon, soot Jes suivants : btct.x'f6c;, £.1tixXT� 1tct.p' &A).wv .. . T� 8Lxixt

377

lettres, apres quoi on placera en regard de ceux-la ceux qu'il ne pergoit pas encore distinctement, et on lui fera voir, par comparaison, qu'il s' agit de la meme lettre dans deux combinaisons differentes en recommengant le parallele entre Jes assemblages qu'il inter rete correctement et ceux qu'il ne sait pas analyser, jusqu'a .ce que les premiers Jui servent de mode!e (paradigme) pour nommer - quelle que soit la lettre et !' assemblage ou elle se trouve - differemment Jes Iettres differentes et toujours identiquement la meme lettre. II suffit d'une confrontation rapide de quelques passages relatifs a l'E7ttx.y(t)y� pour discerner les divers elements, conceptuels et terminologi­ ques, qui la constituent ou Jui sont attaches dans la pensee platonicien­ ne. On notera d' abord !'importance capitale de la perception sensible dans ce type d'apprentissage. On fait travailler l'eleve sur des objets concrets26• Ce n'est pas un hasard si tout le passage est une reflexion sur la notion d'exemple (7t<Xpoc3e.yµ<X). D'ou la necessite de provoquer chez I'eleve une prise de conscience de ce qu'est un exemple, un modele. L'enseignement consiste en premier lieu a donner a l'apprenti Jes moyens d'analyser en Jui un 7t&:6o<;27, et cela en Jui revelant un donne (inne ou deja acquis), qu'il ignorait j usque-la avoir en sa possession, qui Jui servira desormais de paradigme - c'est-3.-dire auquel il se referera pour analyser, comparativement, Jes nouveaux objets28. Un apprentissage a travers le sensible, Jes images sensibles et Jes perceptions sensibles29• L'enseignement se fonde principalement sur la perception et a pour intention d'augmenter et d'affiner l a perception des chases. En accordant cette prerogative aux perceptions, Platon ne signifie pas qu'il rejette .de l'apprentissage le raisonnement : simplement, ii Jui donne pour objet, non point des realites abstraites, propositions ou concepts, appartenant a l ' ordre du discours, mais des perceptions - dont la vertu sera d'etre, grace a Jui, le plus concretes possible. Le raisonnement sert, ici, a ameliorer en l'exeri:;ant la capacite de percevoir Jes chases, et de ce fait, ii apparalt subordonne aux fins de la sensibilite - ce qui pourrait sembler etonnant chez Platon. L' exercice rationnel

�

(26) Le vocabulaire utilise par Platon pour exposer la methode « epagogiq uei> s'appliquerait aisement au domaine musical (8tct.ta6&:ve:o-0ixt 8tcia-oiµci., auAAct.0�). Les lettres sont per�ues comme des sons avant de I'etre comme des signes abstraits, et la <(syllabe une combinaison de sons, un accord au sens technique du terme (U l'origine, la auAAetO�i> est est, en musique, !'accord de quarte). (27) Phidre, 250 8 ; Polihque, 277 D. (28) Tout cela n'est pas sans suggerer quelque rapprochement avec la theorie de la Forme (Geslalt) : inventer un paradigme revient a decouvrir une « honne forme», une forme <( pregnante ». (29) dtO: 't"'ijc; £.vct.pye:a-.&:nic; ci.ta6�ae:w.;, £.vct.pyS.; e:t8wAov (Phedre, 250 D) ; le choix des verbes (ytyvWaxe:tv, £.v8e:tx\IUvetL, 8e:Lx0TI, 8e:Lx0Sv't"ct. : Politique, 278 A-B) est egalement revelateur du fail : Platon a dans !'esprit un type de connaissance essentiellement fonde sur la sensation.

· I .

' 1• '· i ' ,

,1

\,

Ii

r: . :, : I

" !

lil '!i

)·1] :

]i

·111 · : "

'i '1

'

:.

!" I

'

379

F. CAUJOLLE-ZASLA WSKY

INTERPRETATION DU SENS ARISTOTELICIEN D'EilArnrH

auquel est soumise la perception consiste en un travail de comparaison entre des perceptions, et de reference de la perception confuse a la perception claire30. Or, au terme de ce travail se produit ce qu'en empruntant la formule de Piaget on pourrait appeler « nne brusque restructuration )) : a force de repeter le meme mouvement de va-et-vient on voit que quelque chose est un modele 31. Mais l'btocywy� platonicienne ne semble pas s'etre limitee a cette application pedagogique.

contraires particuliers concrets, qu'on voit proceder l'un de l'autre : dormir procede d'etre eveille et inversement, ce qui est plus grand procede de ce qui est plus petit et inversement). Si l'on en croit done Diogene Laerce, Platon aurait fait de l'e7tocywy'ij un usage massif dans ses demonstration (1hoad�e•,)''.

378

Nous disposons, en effet, concernant l'usage de l'€:7ta.ywy-fi par Platon, un temoignage fort interessant de Diogene Laerce, aux §§ 53-55 du Livre I I I des Vies. L'interet du passage tient principalement au fait que !'auteur, attribuant, et a juste titre, a Platon !'usage d'un procede methodique qui se nomme E:7ta.ycuy}j, place sous cette denomination, non point le procede pedagogique que Platon lui-meme a nomme e7tocywy�, mais un certain precede logique qui, certes, a pu etre utilise par Platon, mais qui ne l'a pas ete, en tout cas, sous le nom d'E:naywyfi32• Ce precede logique, ainsi decrit par Diogene « Ea't'L µE:v yap €:7to:ywy� A6yoc, 3LOC 't'Lvwv &A"tj600v -rO Oµ.oLov Eocu't'cf> &.A1J6E:c, olxelwi; E7t�cpE:pwv», possede, d'apres lui, l� trait particulier de pouvoir servir, suivant le tour (Tp67to,) qu'on lu1 donne, aussi bien a la demonstration d'une these qu'a sa refutation. 11 existe, en effet, deux modes d'utilisation de l'e7t�Y"'Y� · L'un procede par contrariete, et se trouve ainsi destine a la refutation des theses adversesaa : mais le second, qui procede de la concordance, est utilise par Platon pour etablir ses propres theses". Ge mode se subdivise lui-meme en deux, comportant en effet une application dite « rhetorique » et qui se limite au champ du particulier3', cependant que l'autre application, (( dialectique », a affaire a I' universe I. Ge mode dialectique procede de la fagon suivante : la these qu'il s'agit d'etablir est presentee sous forme interrogative (par exemple : 1'3.me est-elle immortelle ?) ; pour la demontrer, on a recours a · une proposition universelle (a savoir : « les contraires procedent des contraires ») : mais cette proposition universelle, pour sa part, a ete etablie a partir de faits particuliers (on la tire de !'observation de

(30) E7tt

7tct:pct:60CAAovTct:c;

T&:c; dx6vct:c; t6vn:c; 6e:Wv'tct:�

(Polilique,

278 B).

...

(Phedre,

250 B) ;

'tt6Evct:t 7tct:pOC, 7tct:pct:'tt6i!:µe:vct:,

{31) Ilct:pct:8dyµct:Toc; y' !a'tt T6'te: yEve:atc;, On:6Tct:V .. . (Polilique, 278C). On notera le rOle important de la repetition (souligne par µExptn:e:p en Polilique, 278 B). Elle parait etre le facteur declenchant de la brusque restructuration de l'experience. {32) Diogene Laerce, d'ailleurs, ne dit pas que Platon ait lui�meme nomme En:ct:ywyfj le procede logique en question. (33) 00 µYiv µovo'tp6n:cp, &AA&: 8txTI, § 53 : 0 'te: xct:'t' Evct:v't(watv, oU n:pOc; 't0 8oyµct:'t(�e:tv, &M&: 7tp0c; "O 8te:AEyxe:�v, § 54. {34) Kct:t 0 ex -rijc; .XxoAou6lct:c;, § 53 ; e:tc; 't"1jv TWv �ct:U't� 8oxoUv'tWV Xct:'tct:O'Xe:u�v, § 55. (35) En ce sens qu'elle donne la preuve particuliere de propositions particulieres.

Mais le commentaire par Aulu-Gelle d'une plaidoierie de Caton nous apporte encore, sur l'E7tr:xywy�, quelques precisions nouvelles37. Caton, defendant les Rhodiens, accuses d'avoir eu !'intention de faire la guerre . aux Roma1ns - mais sans l'avoir faite, neanmoins - aurait eu recours dans son argumentation au procede en question. La description de ce procede par Aulu-Gelle le donne explicitement pour un argument peu honorable et trop ose, fourbe, fallacieux : « des habiletes de sophiste grec ». L'argument de Caton est qu'on ne juge pas les gens sur des intentions et, dit Aulu-Gelle, ii aurait « amene )) (induxisse) cet argument par ce que les dialecticiens appellent e7t�ywy�, chose insidieuse, captieuse et sophistique, moins faite pour etablir la verite que pour tromper les auditeurs. Ainsi, Caton s'est efforce, par des « exemples trompeurs », (exemplis decipienlibus) d'arriver a une conclusion (conligere), et d'etablir (confirmareque) sa these. Son argumentation a ete la suivante. S'adres­ sant a titre individuel aux assistants, Caton demande a chacun de s'interroger : chacun de vous, en ce qui le concerne, accepterait-il d'etre puni pour ce qu'il a uoulu faire de mal, sans le realiser? Personne, en fait, ne l'accepterait. Moi, Caton, en tout cas, je m'y refuserais. (Notons qu'ici, le cas particulier, c'est chaque personne presente et a laquelle l 'orateur s'adresse en particulier : « toi, et toi, et toi, accepteriez-vous ? . . . Moi, pas».) Ensuite, Caton imagine un tarif des peines qu'on pourrait encourir pour avoir voulu posseder plus, par exemple (ici le cas particulier touche le contenu particulier du desir choisi comme exemple par Caton : s'enrichir) : on peut etre puni pour etre passe aux actes et etre alle contre la loi pour s'enrichir, mais pas pour avoir simplement desire s'accroitre. Notons que si le procede offre la possibilite d'un usage sophistique, c'est sans doute ici : dans le choix, par l'orateur, des cas particuliers qu'il va mettre en avant (comme : vouloir s'accroitre) auxquels il finira par subst1tuer, comme leur etant equivalent, le cas particulier qui interesse proprement sa plaidoierie (vouloir faire la guerre).

(36) Voir notre article (issu d'un expose presente, en 1987, au seminaire du professeur P. Aubenque, dans le cadre du Centre de Recherches sur la Pensee Antique) : �< Note sur l' Epagoge dans le Sophisle. A propos de Diogene Laerce I I I 53-55f>, a paraitre dans l'ouvrage collectif Eludes sur le Sophisle de Plalon, sous la direction de P. Aubenque, Naples : Bibliopolis. (37} VI iii 34-47. Le terme €7tct:ywyfj figure, en grec, aux paragraphes 35 et 44.

T F. CAUJOLLE-ZASLA WSKY

INTERPRETATION DU SENS ARISTOTELICIEN D'ETIAr.QrH

Apres quoi, Caton a recours a }'argument par les contraires : on ne recompense pas quelqu'un sur ses intentions ; alors il n'y a pas de raison, non plus, de le punir sur ses intentions. D'ou il passe au point de vue general, qui sert de premiere conclusion, qu'il ne faut pas punir les gens sur leurs intentions non realisees. Or, cette conclusion s'applique, en la circonstance particuliere, aux Rhodiens. D'ou la conclusion definitive qu'il n'y a pas lieu de les punir. Voici, maintenant, aussi eclairante1 la critique par Tiroll de cette plaidoierie : les trois arguments de Caton n'ont pas entre eux de commune mesure (quin paria et consimilia non sini). « Vouloir plus», s'agissant de cinq cents arpents est tres different, en effet, de vouloir faire une guerre injuste et impie ; il ne faut pas confondre ce qui est prohibe, illegal avec ce qui est injuste et impie. Et le cas de la recompense est, aussi, different de celui de la punition. En effet, attendre qu'un crime ait ete commis pour le punir, c'est folie pure : ii faut prevenir le crime au lieu d 'attendre que !'irreparable soit accompli. Son reproche general est done que les occurrences pretenduement particulieres n1ont pas ete, en fait, suffisamment particularisees. Caton a procede par amalgames successifs (entre « vouloir plus» et « vouloir la guerre », entre (l illegal )) et (l immoral», entre le principe de la recompense et le principe de la punition). Nous pouvons des lors determiner a conirario, sur ce temoignage, un critere au mains de ce que serait une €7tocywy� non sophistique : ii faudrait, en tout cas, que les exemples (les cas particuliers) soient « paria et consimilia », tires « de chases egales et semblables» (§ 40), c'est-a-dire de meme nature, et qu'on puisse reellement les comparer ; il ne faut pas qu'il y ait « disparililas conlaiionis », « inegalite de la comparaison » (§ 47). 11 ne faut pas trailer comme homogene le disparate. L'usage sophistique consiste a confondre les chases, a les assimiler (ou a les distinguer) anarchiquement, sans respect de leurs veritables points communs et de leurs differences essentielles. Cela, parce que le sophiste se garde de poser, au depart, le critere de ce qui sera dit
Notons qu'un emploi pejoratif du terme semble se profiler deja dans �n texte d'Anstote.' ou il est question, j ustement, de Platon qui, dans 1 •pp i as 3; ura1t fa1t une argumentation captieuse fo dee sur le procede � . ; '""Y"'n . Or, ce que fa1t Platon dans le passage mcnmme, c'est, si 1 on se place dans la perspective d'Aristote, une erreur de genre : Platon p ace dans )e me '.'1e genre OU plutot dans le meme concept genera) (ce)ui '. . d un etre mforme et conscient) l'homme veridique et le menteur comme s'ils en constituaient les differences : mais les differences reelles �eraient en l'occu rre � ce, (au sens d e « qui aime et choisit le faux pour !'amour du faux»), il se , trouve qu on sort du concept sus-mentionne.

380

(38) Bien qu'il soit possible de rendre sophistique n'importe quelle forme de raisonnement, on comprend que le procede que nous etudions ait pu etre juge specialement susceptible d'un usage captieux en raison du rOle qu'y tient !'appreciation du «semblable». Quand cette notion n'est pas strictement definie et qu'on l 'abandonne au jugement de chacun, elle donne lieu a I' expression de la subjectivite la plus totale ; de plus, Jes difficultes rencontrees par Jes premieres tentatives taxinomiques en biologie ont suffisamment montre qu'un critere apparemment evident (comme la taille, par exemple, ou le contour general)

,' ,: I :I �i i'

381

!: � !

Essayons, maintenant, avant de poursuivre, de recapituler Ies . not10ns et les themes qui reapparaissent chez tous les auteurs, dans tous les passages ou il est question de n""Y"'Y� · Le procede est toujours lie a !'analyse de cas particuliers, concrets. La perception, !'evidence sensible, y jouent un role de premier plan ' par tout un aspect n""Y"'Y� se fonde sur !'analyse d'un 7toc6o<;. 11 fait progresser �e raison ? ement a travers le sensible, les images sensibles et les percept10ns sens1bles. Liee a !'importance du cas particulier dans ce type de ra1sonnement, on notera !'importance de la notion d'exemple (""poc3e'.yµ0<), qui apparait comme capitale. Notons aussi le recours exige a ,1., �xper1 �nce I �d1v1duelle, la n cessaire ref�rence 3. un donne (inne OU . deja acqms) qm sert de pomt d appm_ pour Juger, par comparaison des cas particuliers nouveaux. Ce sont des tra its qui rendent le procecte, simultanement, apte a . condmre un apprentissage et fragile devant la sophistique : par sa nature meme, 11 se prete a la casuistique, et s'il n'est pas strictement regle (�om �e d ans le cas de l' enseignement), il fait a la sensibilite (a la . . . de 1 , aud1teur recept1v1te) et, de fagon concomitante, a l'autorite de l'orateur (du « conducteur», de celui qui « mime vers ») une place trap grande. C'est un procede fondamentalement pedagogique (comme le souligne encore le role qu'y jouent la repetition, et le moment-cle de la restructuration, avec les deviations inevitables que cela implique du cote de !'art oratoire). Mainte nant, pour faire sans sophisme un raisonnement par €7tocywy�, . _ Il faut apphquer le procede a des chases apparentees (qu'on pense, chez

�

peut etre un critere de ressemblance tres superficiel et qui ne fait que masquer Jes traits de ressemblance profonds. Metaphysique, .6., 29, 1025a6-13 : -roUTo 8E: 1'eU8oi; Aaµ6&:vet 8tiX -riji; €naywyiji; . (39) (hgnes 9-10). Le passage de Platon auquel ii est fait allusion est celui de l'Hippias Mineur' 365-369.

I!

382

'

I

,,

F. CAUJOLLE-ZASLAWSKY

le, et ii Platon, a l'exem ple des lettres). La notion « TO bµoTov » est capita eclaire Cela cas40• des geneite J'homo de er s'assur est imperatif de e : la proced du if refutat ent egalem J'usage te, contras d'ailleu rs, par refutation opere par rejet du disparate. II faut mentionner encore, dans la constellation de notions qui et le gravitent autour de J'e7tocywyii, celle qui s'exprime par le verbe ces ns, Stolc1e les Chez . substantif correspondant : E:7ticp&pe:iv, &:7ticpopOC le ant ils ton, Pia hez mais � termes signifient « conclure », « conclu sion», . _av1s, non notre a part1e, une sens « appliq uer», « application ». C'est Jes cas negligeable de l'btocywy-lj, et qui signifierait a Ja fois, pour . Y tqu'1Js l genera s � c un sur quer s'appli particuliers, qu'ils viennent , . ue apphq s sel umver I que et Im), par e absorb entrent, qu'ils sont comme mter­ a des occurrences particulieres (theoreme). L'idee d 'apphcat10n chose une sur nom d'un ation }'applic : fagon autre vient encore d'une �t en vo on , encore biais u 1 nouvea ce Par . tesl4 de ve, l'epreu sert de mise a ion : refutat quoi l'E7tocyooy� peut aussi bien servir la demonstration que la un non ou est ]'application sert a montrer si l'objet particulier considere Oµoi:ov, n pas n'est � Oµo'i.'ov de l'universel qui sert de pierre d\�preuve ; s'il t une I'E7tocywy� devient une refutation ; s'il en est un, elle dev1en demonstration. de Munis de J'indica tion de ces caracteres a la fois epars et dotes ne. elicien aristot ctive perspe la a revenir de s coherence, tenton Une chose particuliere nous frappe dans Jes document que nous dans la avons parcourus : dans les textes_ post-aristoteliciens (comm e rsel l'unive conception moderne de l'induction), ii semble aller de soi que en c'est et selle, auquel donne acces l'e7tocywy� est une proposition univer par t, rendan en te, le presupposant qu'on traduit Jes textes d 'Aristo E<po8oi; exemp le, la phrase E:7tixywy� 1j &.7t0 -rWv xix6' Exixa-rov E:7tt -rOC xix66Aou 'pour uels individ cas des (105al3 -14) par « le procede qui consiste a partir loin est l 1t1onne propos re caracte acceder aux enonces universels ». Or ce va ce !'eviden que semble nous ii d'etre eviden t chez Aristote, et meme de tenter allons nous que ce cas tout juste en Bens contraire. C'est en montrer. II y a dans Ja Melaphysique, en M, 4, 1 078b27-3 1 , un texte des plus altre eloquents a cet egard. Aristote a ffirme Ja qu'il serait juste de . recon� » 1que epagog « ement ra1sonn le : ertes decouv deux de a Socrate le merite ainsi que le geste de definir universellement. e: Or ce temoignage jette une lumiere inattendue sur notre problem d'une devient e mis a part le fait que ]'intention pedagogique du proced

(40) Sur 5µ.owi:;, K. v. Fritz [2] p. 202, mentionne !'article « Xenophane» dans PAULY­ W1sso,vA Realencyclop.

1I I

INTERPRETATION

DU

SENS ARISTOTELICIEN o•EnArnrH

383

evidence lumineuse, nous comprenons que Socrate, quant a lui, n 1 a pas invente l'e7tocywy-lj pour fonder la validite de propositions universelles, mais pour arriver a la definition generale d'une notion : l'E:7tixywy� socratique est le passage des individus a la determination d'un concept". L'hypothese est alors permise qu'il en aille de meme chez Aristote egalement. On notera, a l'appui de cette hypothese, que tous les passages de Ja Metaphysique, en l'occurrence, oil figure le terme E:7tet:ywy�, et qui sont impenetrables tant qu'on veut Jes comprendre comme expliquant un passage d'enonces particuliers a un enonce general43, s'eclairent imme­ diatement des qu'on Jes pense en termes de notion generale ou de genre, et d'incJusion OU exclusion par rapport a Ce support (qui peut etre Un concept, un genre, un schema general, voire une regle, mais pas en tout cas l'enonce general d'une proposition vraie). C'est un nouveau biais encore par lequel nous comprenons pourquoi le procede peut aussi bien servir, comme le rapporte Diogene Laerce, a la verification positive qu'a la refutation, suivant que (( l'individu » considere (le cas particulier, ]'element particulier, etc.) entre ou non dans un ensemble donne. Le critere pour en juger etant le couple du meme et de l'autre (eu egard au genre). �

�

(42) Nous avons un autre temoignage de l'origine socratique du procede dans les exemples qu'en donne XENOPHON dans les Memorables (voir, notarnment, IV, 6, 13-15). Aristote lui-meme donne d'autres precisions sur le procede socratique et, en particulier, sur son application exclusive au domaine des notions morales en Metaphysique A, 6, 987bl-4, ainsi que sur le fait que, tout en cherchant !'universe! dans ses definitions, Socrate a su ne pas separer l'universel de l'individu, op. cil., M, 9, 1086b2-13. (43) II est evident que, si €7t«ywyfi signifie
i

f iI 1

',1 .f ; !i

'

1· !.

.;]

.:1

' '

.

:

384

F. CAUJOLLE-ZASLAWSKY

Nous en arrivons, a ce stade de notre recherche, a la conclusion que I'E7tcxywyfi est, pour ainsi dire, la route « au » genre, au double �ens .de : la voie qui mene au genre ou la marche « sur)> le ge nre (en d1rec�1on du _ genre), mais aussi Ja methode qui passe, pour exphquer ou ense1gner .Je comportement d'une notion, par !'intuition - non pas au sens de pressentiment, mais de claire vision intellectuelle - du genre, et plus precisement du principe qui le commande Oe pornt de ressemblance qm definit J'appartenance au genre); et des hm1 �.es hors desquelles. o� sort du . genre (!'extension du genre est donnee par I rnd1cat10n de sa difference la . plus grande : ]'opposition des contra1res). . Le premier sens semble correspondre a l'usage proprement socr� ti­ que du terme. Le « passage 1> qui ca ract� rise 1' �7tocyw .-fi s�g.n1fie a <( menee », ' . a la defin1t10 par la 1na'ieutique, a l'universel, c .est-a-dire n d un co� ce�t (avec ce que le terme de « mener» implique de pedagog1..que). II s �git, sous la direction d 'un (( orienteur» qui conduit le dialogue, d �ne . inlroduclion des participants a la gimeralile en partant de cas particuhers concrets. Le procede consiste, en effet, a prendre appm su� des occurrences particulieres : c'est a partir, pa� exemple, d'une plurah�e. d � , _ choses particulieres di verses, ma1s tou �es d1tes « bonnes �> qu on acce� e a la notion de . Le recours a la compara1son et a Ja ressemblance est done capital, car ii faut percevoir ce qu'ont de commun ces cas particuliers du « bon » : le bon stratege, la bonne cu1ller de bois 1 le bon cheval l'homme bon, et tout le reste de ce qu'on dit tel. Le second sens (� ethode, non de demonstration, mais, si l'on peut dire de « monstration » OU de mise en evidence par application du genre) app � rait plutot lie a l'usage aristotelicien . Toutefois, l'evolution en c� sens de l'E7to:ywy� pourrait avoir commence deja chez Platon, cet emplo1 . paraissant resulter directement de celui de Socrate : une foIS le g;nre trouve par I'emxywy� socratique, ii dev1ent possible de mon:rer (et meme, selon le temoignage de Diogene Laerce sur Platon, de demontr�r) P? r «application du genre», c'est-a-dire en montrant que la notwn etud1ee . . appartient a tel genre donne (btocywy� pos1t1ve) o u ne Im appart1ent pas . . (btocywy� de refutation). C'est la notwn de s1m1htude que semble avoir . . principalement approfondie Aristote. Car le genre en provient directe­ ment : ii repose sur le trait fondamental de ressembl �nce qm fonde une , lignee - et tout en distinguant Jes sens (Metaphys � que 28 : yevo,), Aristote sait fort bien que son « genre » est une hgnee d etres.

�

.�·

L'usage aristotelicien de I'btocywy� se fonde, en effet, sur l'idee centrale qu'il y a des chases qui, par ressemblance, « vont ensemble i> _ (l'alphabet ou Jes nombres en sont l'image la plus frappante), et 11 importe peu que Ieur Iignee soit finie ou infinie : I'impo�tant e�t qu' �lles , se comportent de la meme maniere, qu elles obe1ssent � la meme ;egle ; . Quand Aristote dit qu'etre un nombre a pour pomt de depart etre I umte (Metaphysique, LI., 6, !016bl 7-18), ii applique la conception du nombre .

INTERPRETATION DU SENS ARISTOTELICIEN

D'EIIArnrH

385

developpe en Metaphysique, I , M et N ; et dont ii ressort qu'on obtient un premier nombre en prenant pour point de depart l'unite et en ajoutant un, et le reste des nomhres en recommengant. Voila la regle 1 voila ce qui definit le genre « nombre ». lnterpretons cela en termes d' btocywy�. Le cadre le plus propice a !'utilisation de cette derniere est, nous l'avons vu, celui de l'apprentissa­ ge. Supposons qu'il s' agisse d'apprendre a compter OU d'apprendre ce qu'est un nombre. La question est de faire comprendre concretement « comment ga marche i> en faisant apparaitre sur des nombres concrets que, d'un nombre a l'autre, ii y a quelque chose qui est, a chaque fois, la meme chose et se repete semblablement (le genre est, precisement, ce que deux chases differentes ont en commun). Voyons alors comment se distribuent Jes divers aspects de l'btocywy� apparus lors de nos precedentes analyses. Les cas pariiculiers sont, bien entendu1 Jes nombres sur lesquels on montre la regle « ajouter un». Et peu importe, a notre avis, la quantite de ces cas particuliers : elle ne depend que de la rapidite de l'apprenti a reperer une constance sous couvert d'une diversite. En l'occurrence, la diversite est celle des dif!erents nombres et la constance est dans la regle de leur succession. Et peu importe1 egalement, que la suite des nomhres soit sans fin (le procede n'a rien a voir avec l'induction complete), quel que soit le nombre particulier considere, on sait - une fois qu'on a vu le comportement du nombre, qu'on a assimile ce qu'est un nombre - qu'il sera constitue de celui qui le precede augmente d'une unite. On peut appeler cela une generalisation, mais a condition de l'entendre litterale­ ment, c' est-3-dire comme le passage au genre, comme l' acquisition du genre et la capacite, derivant de cette acquisition, d'en faire usage - ce qu'on fait en l'appliquani a de nouveaux cas particuliers, c'est-3.-dire, en }'occurrence, en appliquant a chaque nouveau nombre la meme regle. Cependant, s'il y avait proprement generalisation, c' est-a-dire passage a des propositions vraies universellement, Aristote n'opposerait sans doute pas l'btocywy� au raisonnement deductif (Myo,), au syllogisme (cru1'1'oytcr­ µ6,), a la demonstration (om63eo�<,). La generalite de l'eitocywy� est celle de l'application r nniverselle » (c'est-a-dire l'application a chaque nouvel element du genre) de la regle qui definit le genre. En un sens, on reste toujours sur le plan des cas particuliers : on ne fait que passer d'un cas particulier a un autre. On ne traite effectivement que du particulier. La notion d'universalite, ou de totalite, n'intervient pas, en effet, dans le precede en question, comme telle, c'est-9-dire a titre propre et distinct, autrement
386

F. CAUJOLLE-ZASLAWSKY

cas doive imperativement entrer en jeu dans une E7tocywy1j) ; et, m€:me s'il est vrai qu'un predicat universe[ est celui qu'on peut dire « de loul», ii suffit de substituer la notion de necessite it celle de ce « tout», pour constater, devant !'indifference d'une telle substitution chez Aristote, que la totalite en question n'a pas de signification qui soit fondamentale­ ment numerique44• Dans l'&7tocywy1j d'Aristote, on peut toujours mettre le nombre de cas entre parentheses. II n'est pas capital. On peut faire une E7tocy(t)y1j ayant valeur « universelle )> sur deux cas. La, c'est-3-dire dans le fait que, pour Aristote, un universe} n'est pas un individu et n'est done pas « separablei>, se trouve, nous a-t-il semble, le point essentiel : Aristotle a vu dans l'btocywy�, it la stricte condition qu'on la pratique dans une perspective socratique, une methode qui, s'agissant de l'universel, a l'incomparable merite - par contraste avec !'usage deviant de Platon qui con�oit !'abstraction comme une separation - de ne jamais disjoindre le genre et l'individu. Et c'est pourquoi Aristote a voulu, par-delit Platon, tenir directement de Socrate sa propre btocywy�45• __

T

- 'I I

I I

I

INTERPRETATION DU SENS ARISTOTELICIEN D'EITAr.O.rH

[7]

T. H. IRWIN,

387

«Aristotle's discovery of Metaphysics )>, The Review of Metaphysics 31 (1977), 210-229. « First principles in Aristotle's Ethics», Midwest Studies in .Philosophy I I I (1978), 252-272 (254 sqq. notamment). [8] T. ENGBERG-PED�RSEN, , Phronesis 24 (1979), 301-319. [9) J. HINTIKKA, <1 Aristotelia n induction l), Revue lnfernafionale de Philosop·hie 1980 , 422439. ' [10] L. CouLo�BARITSIS, « Y a-t-il une intuition des principes chez Aristote ? », Revue lnternaltonale de Philosophie, 1980, 440-471. [1 1] T.V. UPTON, «A note on Aristotelian epag&ge», Phronesis 26 (1981), 172-176. [12] S. BEsor.r, « Percezione, verita e giudizio : Luoghi dell' intuizionismo aristotelico» ' Ann. Disc. filos. Univ. Bologna, 4 (1982-3) 5-40. '

INDICATIONS BIBLIOGRAPHIQUES

EVANS, « Causality & explanation� in the Logic of Aristotle», Philosophy & Phenomenological Research, 19 (1958-9), 466-485 (477 sq.). K. VON FRITZ , « Die epagoge bei Aristoteles», SB d. Bayer. Akad. d. Wiss., Philos.-hisl. Kl., Jg. 1964, H. 3, Munich, 1964, reprod. clans Grundprobleme der Geschichle der aniiken Wissenschaft, Berlin, 1971, 623-676. W. HESS, « Erfahrung und Intuition bei Aristotles)>, Phronesis 1 5 (1970), 48-82. D. W. HAMLYN, «Aristotelian Epagoge », Phronesis 21 (1976), 1 67-184. L. A. KosMAN, « Understanding, Explanation and Insight in Aristotle's Poslerior Analylics», dans E. N . Lee/ A. P. D. Mourelatos / R. M. Rorty eds., Exegesis & Arguments, Studies in Greek Philosophy presenled lo Gregory Vlaslos, Assen : Van Gorcum (Phronesis, Suppl. Vol. I), 1973. , W. N . THOMSON, Aristotle's Deduclion and lnduclion, Amsterdam : Rodopi, 1975.

[1] M. [2] [3] [4] [5]

[6]

(44) L'equivalence, chez Aristote, entre necessite et quantificateur universe! a ete notee par J. LUKASIEWICZ, dans son ouvrage Aristotle's Syllogistic, from the standpoint of modern formal logic, Oxford, 19572 (1951) § 5 (p. 30-31 de la traduction frangaise, Paris : A. Colin, 1972). (45) Le passage, deja cite plus haut, de la Metaphysique M, 9, 1086b2sqq est suffisamment explicite sur ce point, qui dit en substance : Socrate, en introduisant les definitions, a eu le merite de ne pas separer !'universe\ de l'individu, et il a eu raison de ne pas Les separer. A juger sur pieces, on voit que, d'un cote, sans l'universel, la science n'est pas possible : mais, d'un autre cote, si !'on separe cet universe!, on tombe dans toutes les difficultes liees a la theorie des Idees.

'

:

1 I

I

B1ot.oora. LOG10uE ET METAPHYS1oul!

"""' """""' 8eminaire CNRS-N.S.F., 1987

fiditions du CNRS, Paris, 1990.

NATURE AND CRAFT IN ARISTOTELIAN TELEOLOGY SARAH BROA D I E

PART A: INTRODUCTION (I) It used quite commonly to be argued against Aristotelian teleology that this type of explanation founders for lack of a suitable psychological basis in most of the phenomena for which it was invoked. The objection was that it does not make sense to attempt to explain X as happening or existing in order that Y should happen or exist unless one assumes that the production of X and Y is guided by thoughts, desires or conscious purposes relating Y as end to X as means. Yet Aristotle himself makes it clear that the domain in which, as he believes, teleological explanation is not merely illuminating but mandatory if we are to have any adequate explanations at all, extends well beyond the domain of entities to which beliefs, desires and conscious purposes can be reasonably ascribed. Nor, on the other hand, does Aristotle protect himself against the objection by introducing a supernatural being who produces or shapes phenomena in accordance with divine intent. Thus if the application of teleology does depend on the presence of psychological factors such as those mentioned, it is a dependence which Aristotle fails outright to notice. Alternatively, there is no such failure because the claimed dependence does not hold. (2) This poses the question: Is a non-psychological teleology as intelligible as Aristotle evidently takes it to be? We may be tempted to turn a deaf ear to this question if it seems to us that in other respects Aristotelian teleological explanations are effective or at any rate more effective than alternatives given the state of knowledge. But embar­ rassing though it may be for Aristotelians, the question is prima facie a pressmg one precisely for them. What makes it so is Aristotle's

1 i

390

S. BROADIE

characteristic appeal to the notion of craft (techne) as analogue for his conception of nature (physis). Nature, in this context, is not Nature in general, or the cosmos, but the specific essential nature of an individual . . substance, the inner principle of its behaviour and orgamsat10n. It ·1s nature in this sense that Aristotle likens to craft - that is, to one or another specific craft; for the crafts are principles of activity vested in particular individuals who live, move and have their being within the physical world. Craft in its active exercise is evidently end_-directed, and to Aristotle the same is true of nature, although less evidently so. Thus it is craft that provides the model for nature, not the reverse. That the comparison goes in this direction might seem too obvious for mention, except that just at this point we confront what seems the most vulnerable spot in Aristotle's position. If one were to establish independently the idea of the nature of a thing as end-directed, and were then to use this conception of nature to explain what should be meant by the end-directedness of craft, one might well achieve the interesting result of having developed a notion of craft (that is, the exercise of craft) in which psychological concepts play no essential part. For generally speaking, natures are not psychological, and in this backwards analogy it is craft that is likened to nature. I shall return later to this possible way of conceiving of craft. Meanwhile, in Aristotle's actual account craft is the model, and the nature of a thing the explanandum: which is not surprising, since we have more of a pre­ reflective grasp of the idea of craft than we do of the idea of a thing's nature. But now the vulnerable side: isn't it also part of that familiar pre-reflective notion of craft that craftsmen are human beings operating from beliefs, desires and conscious purposes - and that without all this craft would not be possible? But if that is so, are we not right to mistrust Aristotle's confidence that nature can be coherently treated as teleological like craft, even though the operations of nature, unlike the operations of craft, mostly do not depend on psychological attitudes? (3) There are a number of problems here. I shall indicate two, with the aim of focusing on the second. First there is the traditional objection to natural teleology: " How can a state of affairs that is at best future and at worst never occurs at all (since the end may not be realised) exert an influence in the present? Only something now can cause something now: for example, a present desire causes a movement. But nothing future can cause a present movement otherwise than as the object of something present; in which case what really causes the movement is the present item, of which the future one is only the object. So take away such present items as desire, will and conscious design, and you take away the basis for applying any explanation that refers to what is not yet." It is sometimes commented that this argument crudely mistakes final for efficient causes: efficient

NATURE AND CRAFT IN ARISTOTELIAN TELEOLOGY

391 causes cannot operate unless as already existent, but the whole point about final causes is that they can (not that this "can" attributes some extraordinary capacity); thus surprise here simply betrays failure to grasp the idea of final causality. But this response is itself inadequ ate msofar as it suggests not only that a final cause is not an efficient cause, but that a final cause can function on its own without an efficien t cause. No doubt that would make sense if one could indeed intelligi bly thmk of a final cause as a surrogate efficient, something planted there in the future and backha ndedly stirring things up in the present. That is what the objection rightly finds to be no nsensical. But in Aristot le, final and efficient causality are complementary: the end is an end of or for an agent, and the agent as such is bent on an end. And this complementarity is general, applying where will or desire is present, but also where they are not. Thus it is not as if the absence in a given case of empirical psychological factors forces us into the following choice: either (a) there is only a final cause functioning in a vacuum (which is absurd), or (b) final causality does not obtain at all, or (c) final causalit y does obtam, but only through the medium of a specially postulated empirically unidentifiable desire or the like. These options are not exhaustive unless one assumes that it is only in virtue of a desire or conscious purpose that an agent can be an end-oriented efficient cause of some objective which in turn functions as final cause of that active efficiency. In other words, the assumption is that intentionality - or more precisely, end-wardness - must be mental or grounded in the mental, in the sense of requiring a mental representation of the end. (4) However, Aristotelian teleology clearly dispenses with this assumption. That brings me to the second problem: Does non­ psychological end-directedness make sense? This paper is not an attempt at a general answer to this questio n; it has the narrower purpose of examining for coherence one exampl e of the view at issue - a surely paradigm example , though, since Aristotle is the greatest historical exponent of this way of thinking. A verdict on coherence requires consideration of the relation between Aristotle's doctrine of natural teleology and his use of the craft analogy . That is my concern in what follows. I shall mainly consider two questions: (i) Does the doctrin e depend on the craft analogy? (ii) Does the craft analogy import undesirable psychological elements into the doctrine? If the answer to the first question is No, then it does not matter if the answer to the second is Yes. I shall, however, argue that the answer to the first is Yes, and to the second No (respectively in Parts B and C). I shall then (Part D) consider further problems arising from the craft analogy.

I. :

392

S. BROADIE

NATURE AND CRAFT IN ARISTOTELIAN TELEOLOGY

PART B: Is THE CRAFT ANALOGY NECESSARY?

matter as explicable in terms of it. Thus we move to saying that it is because there are, and eternally are to be, such organised living systems that the materials which form their bodies, their food and their environment, arrange themselves in ways that make it true that those systems are and are to be. In other words: since they are and are to be, and since this cannot happen unless the materials behave in certain ways, the materials must behave in those ways - in order that the forms be realised.

(5) An excellent starting point is already at hand in John Cooper's paper "Aristotle on natural teleology" .1 By a very clear and thorough argument Cooper shows a way in which we may understand Aristotelian teleology without tacit appeal to psychological factors. The argument rests on Aristotle's view that the species are eternal. Cooper presents it as a fundamental fact of the Aristotelian universe that the organic forms realised at any given time should never not be realised through indiviJuals of the relevant kinds. This fact, on Cooper's account, is what we would probably call an ultimate law of nature. That it obtains is not something that can be explained. In particular, it cannot be explained as due to the movements and properties of the inorganic materials to be found in the universe. Cooper stresses, as others have done,2 that given the evidence available to Aristotle it would be entirely reasonable to conclude that the behaviour of inorganic materials could not account for the formation, preservation and propagation in saecula saeculorum according to kind, of those highly organised self-maintaining systems we call living creatures. On the other hand, to put all this down to sheer accident is simply absurd. Moreover, even if (as is assumed not to be the case) organic phenomena could be explained materialistically in terms of the coming together of inorganic elements each pursuing its own natural course, the biological viability of the resultant wholes would still be an accidental by-product of the workings of those simpler natures. For what would be explained would only be the individual positionings of the elements which, positioned as each is, happen to compose viable organic unities. The unities as such would not be explained. And that, for Aristotle, is no less absurd, given that these supposed ' by-products' are, in his view, eternally reinstantiated. But the only alternative to treating the. eternal re­ instantiation as a brute fact unencompassed by any explanation is to treat it, itself, as a basic principle of explanation. So instead of thinking of it as belonging on the level of what it makes sense to try to explain in terms of matter - or to bemoan as inexplicable if such explanation fails to fit - we should rather think of the behaviour of

(1) In Language and Logos, Studies in Ancient Greek Philosophy, edd. Malcolm Schofield and Martha Nussbaum, C.U.P. 1982. (2) E.g. Allan Gotthelf, "Aristotle's Conception of Final Causality", Review of Melaphysics 1976; Sarah Waterlow, Nature, Change, and Agency in Aristotle's Physics (O.U.P. 1982), eh. 2.

393

(6) This analysis is broadly in the spirit of the texts,3 and, as Cooper points out,< it has the merit of showing how a rational teleology need owe nothing to psychological assumptions. Granted that under certain intellectual conditions it is only reasonable to apply teleological explan ation to organic phenomena, then no[ to apply it, '(given such . cond1t10ns, would be (one might argue) so unacceptable that even if the psychological assumptions were legitimately available their addition could hardly strengthen the already adequate case for teleology. But now how does Aristotle's craft analogy fare in the light of all this? On Cooper's account it seems inessential. Thus Cooper writes: " . . . one must reject the suggestion that is sometimes made that this analogy is central and fundamental to Aristotelian teleology. "5 The dismissal is surprising. For surely it is Aristotle's own texts that cumulatively make the suggestion through innumerable emphatic occurrences of the analogy. Cooper correctly points out that at least one of Aristotle's arguments for natural teleology does not rely on the analogy.' However, it does not follow that Aristotle's full view can be properly represented without it. It is possible, of course, that he gives the analogy more prominence than it deserves for the amount of good it does; but why should one suspect this? One motive, if not reason, that might weigh with an apologist for Aristotelian teleology would be the thought that the craft analogy carries unwanted psychological implica­ tions. (7) Unwanted implications or not, I shall now argue that the craft analogy, though possibly dispensable to some versions of natural teleology, makes an essential contribution to the Aristotelian version. This contribution is, however, of metaphysical rather than scientific significance. Hence it may well be overlooked in the context of the debate on the scientific merits of teleological as opposed, say, to mechanistic explanation. Aristotelian teleology, as clarified by Cooper, rests on two presumptions whose reasonableness depends on empirical

(3) But for an important qualification see paras. 8-9 and note Cooper, op. cil., p. 221. Coopec, p. 198, fn. 2. (6) Phys. 19�b32-l 99a8.

(4) (5)

I

7.

..

'

'

I i

S. BROADIE

NATURE AND CRAFT IN ARISTOTELIAN TELEOLOGY

evidence together with certain extrapolations from this. One is that the living forms are never not instantiated; the other that this fact cannot be explained in terms of the material components. There is also the consideration that in biology, teleological explanations are found to work well; for example, they generate predictions, and the predictions are often confirmed. Now the empirical facts (or presumed facts) that make teleology scientifically attractive do not in themselves dictate any underlying metaphysic. Indeed, there is no reason why the _teleolog1sing scientist, any more than his mechanising conterpart in some other . age (or other field of phenomena), should embrace any particular metaphysics at all. To give a non-teleological example: the reactwns of a compound at various temperatures are explained in terms of molecular theory. That explanation is consistent with distinct metaphysical positions regarding the ground of what are empiricaBy identified as causal connections. Thus one may suppose a non-empmcal bond or tie by which a molecular re-arrangement is directly 'powered' by a change in temperature; or one might take the sequent phenomena to be each 'powered' to occur in the sequence by the will of God alone. Or one might hold that nothing whatever is the case beyond the fact that, this time too, an event of the first type is followed by one of the second. (8) Natural teleology is likewise open to a variety of metaphysical interpretations. For example, one might hold that it is inherent in the nature of something called "the Universe" that certain forms be always realised. On that view, the form-realising behaviour of materials has a single metaphysical ground, namely the nature of the Universe. I f the realising of some form is to be considered a goal, then it is a goal for the Universe, an imperative incumbent upon the cosmos.' It is not a goal for any of the circumscribed physical things whose behav10ur and development is teleologically explained as being necessary for the realisation of the form. For if the goal is the Universe's goal, these physical things are its means and instruments; but means and instruments do not 'have' a goal in the sense in which what uses them has. To make the point we need not suppose the Universe to be a thinking being with conscious purposes, any more than we need suppose a transcendent God who intends that the forms be realised and consequently wills that matter falls into the necessary patterns. The

point is threefold: firstly, that the conditions under which it is appropriate to apply teleological explanation as analysed by Co.aper may be satisfied even on the assumption that it belongs to the nature of the Universe that organic forms be always realised; secondly, that such an assumption is thoroughly unaristotelian; thirdly, that it fails to be Aristotelian not because the Universe is conceived as a conscious being (this is not necessary to the assumption), but because there is no place here for Aristotelian natural substances. According to the picture just sketched, the realisation of some given form is a goal served by but not grounded in the natures of whatever physical objects behave in the appropriate ways. Are we now even entitled to think of these as having natures at all, in Aristotle's sense? For what he means by a "nature" is not merely a set of dispositions to behave in certain ways, but the inner source of such behaviour.' In particular, it is the source of the changes by which relatively undifferentiated matter develops into a given form. To speak of the physical thing as the source of its changes in this way is to say more than that it exhibits change that cannot be accounted for by physically external factors. That is an empirical matter, and is compatible with the metaphysical view that all changes are grounded in the nature of the Universe. The Universe thereby becomes the one substance, according to Aristotle's equation of being a substance with being a source of change.' Ordinary physical objects, on this view, must be construed as modes of that substance, they being merely subjects of change and not metaphysically independent sources. (9) The difference between this and the metaphysical teleology of Aristotle may be illustrated by reference to the difference between two ways of representing goals, one predicative, one propositional. It is common ground that (let us suppose) relatively undistinguished blobs of matter (e.g. frog spawn) change and develop in ways which one cannot begin to explain in terms solely of what is present - the shape, weight, viscosity, temperature, etc. - but which fall into place in the light of our knowledge that what each blob will eventually have become is a frog. Should we then say that the goal is that there should be (or have become) a frog; or is it rather lo be (or have become) one? The former represents an end logically appropriate for the Universe as agent; the latter for an Aristotelian natural substance, which in this example is precisely that potential frog on the way to becoming what it is its goal to become, namely an actual one: which actuality, if achieved, is achieved not merely for it but by it. Now this is the conception that the craft analogy is surely intended to hammer home: the conception, namely, of

394 ' •, , ,

:\,' ' !

·'

I 1·

' .: : · i

i ;I

'I: ,:

(7) Cooper's language frequently leans in this direction; e.g. : " ...it is an inherent, non-derivative fact about the natural world that it consists in part of natural kinds and works lo maintain them permanently in existence ... it is a fundamental fact about lhe world . . that it maintains for ever these good life forms ... " (p. 213; my emphasis). But, strictly, Aristotle's metaphysical pluralism implies that there are as many 'fundamental facts' as there are distinct kinds of organic substance: for each kind the fact that ii is always instantiated. .

(8) Phys. 192bl3-23. (9) Phys. 192p33; cf. 1 93a9-IO and 20.

395

T I

396

, ·1 r. [:.

f!!' ..' I ,

·: '"

S. BROADIE

particular physical things as themselves metaphysical centres of the development towards form. For the craftsman too is a source of change operating as one of many within the order of nature. The difference is that his goal is not to have become a so and so, but to have made something into a so and so by the practice of his art. So far as natural objects are concerned, one might easily suppose them to be mere subjects of change, who knows what the source of it? But with craft, the functions of source and subject are clearly partitioned, since in general the development for which craft is responsible is loca ted in an object external to the craftsman.30 By taking craft as model for the specific natures of particular physical objects, Aristotle unambiguously declares their status as metaphysical centres of activity. Without the craft analogy we should still have teleological explanation, but not the Aristotelian concept of natural substances. PART C: A NON-PROBLEM OF THE CRAFT ANALOGY (10) It would be unfortunate if the craft analogy, necessary as I have argued it to be, should finally prove dependent on psychological assumptions alien to Aristotelian teleology. This is now the question. Let us start with the consideration that this analogy is a complex package containing very much more than the elements of belief and desire sometimes deemed necessary for teleology to make sense. I f these are what has to b e invoked to.support the notions of goal-directed natural processes, why say more than that such processes are like the operations of a consciously purposeful agent? Why confine attention to craft, when so many of our deliberate ends and means owe nothing to special expertise? Why not take going to a neighbour's house in order to find out the latest news? Here are some reasons. (a) The world of craft is divided into craftsmen of various kinds, each qua craftsman seriously dedicated to his own speciality. So it is with the world of natural substances. For each individual there is the circumscribed end proper to its specific definable essence, and for each an equivalently circumscribed range of means. There are no amateurs here, or dilettanti. (b) The matchless professionalism of healthy_ natural subs­ tances is shown in the regularity of their behaviour. Craft, with its rules and repeatable applications, echoes this, even though a realistic view of human history must allow that crafts develop, discovering not only new methods but new horizons. (c) Nature makes correct moves, and so does craft. That is to say: while a craftsman may make mistakes, his claim

(JO) Phys. 192b15-20; Melaph. 1070a7.

NATURE' AND CRAFT IN ARISTOTELIAN TELEOLOGY

397

to the title, as Thrasymachus pointed out in the Republic,11 depends on knowledge, not on misjudgment. Those who think that teleological explanation needs psychological assumptions are concerned with beliefs (as well as with desires, intentions, conscious purposes). But to play the desired explanatory role, the beliefs need not be correct. It can hardly be that the image of the craftsman is invoked mainly in order to supply the element of belief. (d) The metaphysical unity of an Aristotelian nature is mirrored by the teleological unity of craft-cognition with craft­ goal. By this I mean that the knowledge exercised in craft is normally developed precisely with a view to the craftsman's purpose. This is t.rue not only of the necessary causal generalisations, but also of the particular observations required for applying these in an actual situation. The cognitive system exists and operates only as called upon for the end. Idealised, this concept approximates to the concept of an Aristotelian nature, where the end and the capacity to realise it are perfectly integrated. It is not as if the capacity, in any given case, could have been used to some other end, or is more than is needed for this. Contrast, in this respect, both nature and craft with what we may call "ordinary" human purposive action, where, as likely as not, the relevant desire or interest, and the relevant beliefs or cognitive states, were developed independently and come together by an external synthesis. (e) 'Ordinary' purposive actions are usually explained by saying that the agent desired so and so, or that he had a reason for pursuing so and so. Such statements do not as a rule appear in explanations of the activities of natural substances, since in general we do not think that these operations express desires or reflective concerns. But the same is true of the operations of craft, considered simply as such. There is a sense in which the craftsman as such is loo practical to be animated by desire or reasoned concern for the end. To say of someone that he operates as a builder is already to have implied that he pursues the builder's typifying end (not merely that he exercises the special skills). The builder as such cannot want to build houses in any sense in which wanting to build houses could explain building by the builder. Nor is it his business, qua builder, to have reasons for building. The builder here is, of course, an abstraction, and it is only to this abstraction that these remarks apply. They do not suggest that a human individual might not engage in builder-activity for some reason, or because he wanted to do this work or have its product. A tree would have to be more than a tree if it were a tree because it desired to be a tree. This is not because trees are incapable of psychological attitudes such as desire, but because even the desires of such organisms, were they

(11) 340c-e.

7

S. BROADIE

398

to have them, would be expressions of their perfect - tree-form seeking tendency, hence not explanatory of it. (1 1 ) Point (e) above is the direct claim that the craftsman as such cannot be said to have the psychological attitude of desire for the craftsman's goal. Some of the previous points should help to construct a similar case regarding other psychological attitudes. Aristotle says that craft does not deliberate. 1 2 He might have added that in paradigm cases the craftsman in action does not even have thoughts about whatever it is he does that constitutes his exercise of skill, except under two circumstances: one, where he is still learning, and the other, where he is demonstrating so as to teach the art. But in the first case he is less than a craftsman, and in the second case more. With regard to the ascription of belief, two features should cause us to hesitate. No doubt we are concerned with an idealised case of craft, but it is in terms of such that imperfect cases are to be understood. First, there are no mistakes; second, there is perfect fit between the cognitive system and the end to be achieved. But now where the question of error does not arise, is there need or room for the notion of belief (at any rate so far as this involves the idea of a true-or-false mental representation)? And where the cognitive material is wholly at the service of a given end, is there room for ascribing a factual (or proposition-asserting) attitude such as the word "belief" commonly implies? To answer Yes with confidence one would have to be sure that our distinction between belief and, say, desire (which distinction plays a .significant part in determining the meaning of "belief") could still be drawn in the case mentioned. But would that be possible if the alleged belief had no life or history of its own independent of that one conative context? (12) These remarks are only pointers, but pointers in a dire�tion which for the present limited purpose is perhaps sufficiently clear. We have at any rate shifted the burden of proof to those who would deny what is here suggested : namely, that craft is non-psychological in precisely those respects in which craft is most suited to provide the model for nature. PART D: FURTHER CONSIDERATIONS (13) I have argued that Aristotle's craft-analogy for nature does not introduce the psychologism sometimes feared. But his use of the analogy is by no means all plain sailing. Here I shall examine some problematic ramifications. In the first place, the concept of craft is (12) Phys. 199b26-8.

T

NATUR E AND CRAFT IN ARISTOTELIAN TELEOLOGY

399

supported by presuppositions which on reflection might lead one to question the propriety of using it to further our understanding of nature. One such presupposition of course is that a natural order is already there, providing conditions for the growth and exercise of human skill. Even if craft is itself an expression of human nature craft is a rabonally organised attempt to control objects and forces alre � dy at work independently. Thus craft is a fitting analogy for the nature of a non-human natural substance only to the extent that we can conceive of such a substance as depending for its orderly development on natural . quahties and relat10ns of natural objects other than itself: which objects, m this context, are regarded as materials for the realisation of the natural form in question. Aristotle finds no trouble in understanding the world of nature in this way. Nor, I think, should he, so far as method is concerned. Someone might object that if non-human natures in general are to be conceived on the model of the craftsman, then the presupposition just mentioned will force us round in a circle when we come to analyse craft. But the objection is superficial: although we cannot make sense of craft without presupposing natures which craft has not touched, there is no incoherence in treating the latter as also craft­ like in turn. For it is not in the same conceptual breath that we both oppose them and liken them to craft. Even if natures are craft-like it is not on account of this resemblance that nature is a precondition for craft. The dependence here considered of craft upon what is not craft can be grasped witho ut circularity as well by Aristotelians as by those . philosophers who claim to find nothing craftsman-like in nature. (14) But there is another presupposition which gives rise to doubts less easy to weigh. Just as craft depends on a world of non-human substances confronted as independently there, so craft assumes non­ craft on the human side as well. Craftsmen must know what nature makes possible given the human will to guide and divert natural processes towards human objectives. But the objectives themselves ultimately rest on principles belonging to no special craft. The craftsman, whatever his kind, aims at some good; but to say this is to say more than that the craftsman aims at the end that defines his craft. For the end aimed at is not good because in that craft it is the end aimed at; rather, it is aimed at because it is of value as answering to some human need, passion or interest. No doubt if the carpenter were nothing but a carpenter, the good as conceived by him would be nothing other than the production of cabinet work; but the same would be true mulalis mulandis, of the craftsman, if such there were, whose metier il was to blow curious arrangements of soap-bubbles. In fact, however, it is good that certain articles are produced, because these articles are regularly wanted for reasons having nothing to do with the craft of their production. This is by contrast with the hypothetical case in which the defining end of some conceivable craft would be good if its products were

T

" 1'

400

S. BROADIE

of any importance. Now the real carpenter, an intelligent human being, operates as such: that is, with an eye to the needs that called forth the exercise of his craft in the first place. While expertise may be necessary to convert, say, the need for shelter and for various utensilS into practicable objectives, one does not have to be an expert to appreciate the desirability of such products. This rests with the judgment of those who are not carpenters, or not merely carpenters. Aristotle, as everyone knows, takes such judgme(lts to be grounded in a conception of human happiness of the good of man qua man. The lowly carpenter may not give this much thought in the sense of concerning himself with some all-round definition; but he responds to market forces, or ignores them at his peril, both with respect to what he makes, and to its timing and quantity. Thus it is only in a highly abstract sense that a given type of craftsman always produces the same. There are times when he produces nothing even though awake and physically capable, and so far as he does produce, the character of his objective is constantly subject to re-interpretation such as never occurs in the non-human world according to Aristotle's view. (15) In short, the specific crafts are not autonomous but operate within limits set by what Aristotle calls "phronesis": wisdom about the whole of practical life. It is true that Aristotle sometimes follows Plato in a schematic division of labour: the carpenter is one entity and the being who dictates the when, where and how much of the carpenter's activity is another - the politikos, _whose concern is for the good in all its aspects.13 But that is for the purpose of analysis only. The actual carpenter has to be something of a politikos: not necessarily a 'statesman' in some high sense, but a social being. His excellence in his particular trade depends in part on this, for who would employ a carpenter who inderstood nothing but carpentry? More to the point, such a notion is hardly intelligible. Craft is essentially practical, or (to speak more abstractly) a source of orderly change. That is the basis of the analogy with nature. But from craft alone nothing comes about. For even if what the craftsman does in making what he makes can be explained by the principles of his craft, that these principles operate at all, and when and to what degree, depends on ulterior values -" This is by contrast with the nature of a natural substance, which is not merely an inner principle of change, but, as Aristotle says, an innate impulse (horme) 15 of change, so that the characteristic developments must occur in the absence of physical impediment. When, as in Melaph. Z 7, 16 Aristotle compares the active

(13) Eth. Nie. 1094a26 ff., for instance. (14} er. ibid. I 139a35-b3. (15) Phys. 192bl8-I9. (16) I032a32-bl .

I

NATURE AND CRAFT IN ARISTOTELIAN TELEOLOGY

401

essence of a natural thing with "the form in the soul of the craftsman" he sp�aks as if the craft-form itself holds all that it needs in order to b � funct10nal. In effect, then, Aristotle here treats craft as conceptually isolable from those non-techmcal factors that set its objective. But this is to distort the not10n of craft, if (as I argue) we cannot make sense of craft otherwise than as practical, and further cannot make sense of craft as actively practical without noting its dependence on something other than itself, namely general practical wisdom. This is not to say that _ thi s isolated not10n of craft may not illustrate something of what _ Aristotle understands by "nature"; but it does mean that he cannot safely model nature on craft, since if we take one thing as model for another, we expect the latter's structure to exhibit all that is essential in the former. If the natures of things really parallelled human crafts then (as has already been indicated) nature would not behave with th � _ massive constancy that is its hallmark. And the parallel undermines _ _ that metaphysical pluralism which, as I suggested in the second section the craftsman analogy helps to sustain. We now see that this help i� afforded only by a notion _of craft plucked out of the system of . conceptual connect10ns that give it hfe. For considered in the round a craft is more like an organ than it is like an independent substanti al principle. With an organ there must be a user who rationally co­ ordma tes the use of it with the use of others. The user of a craft is the _ . �ocial mdividual, who stands in this relation to many crafts, although his mvol�ement with them takes different forms depending on which are practised by him and which by others. Once this line of thought is _ apphed to nature, it becomes tempting to postulate a single universal _ _ _ natural prmciple directed towards some one end to which each part of the physical world must be understood as subordinate. In a few _ passages Aristotle seems to move in this direction. We need not w�nder, considering that the craft-analogy, unguarded, is bound to lead this way. (16) I have attended so far to one point at which the concept of _ _ _ craft must be circumscribed if 1t is to play its expected part in the _ analogy: this was at the mterface of craft with practical wisdom. But there are further necessary restrictions, of which some have to do with the specialist aspect of craft. Crafts are learnt, and they are passed on _ which _ respect the species of craft differ essentially by teachmg: m from the species of nature. For it is surely of the essence of craft to be transmitted by cultural as distinct from genetic inheritance. Here, perhaps, is the ground for that special dimension of rationality that . _ makes craft_ so mtensely s1gmficant to the Greek philosophers. This is _ th� dimens10n m which functioning in accordance with rational prmciples is mtegrated with the reflective knowledge of those principles _ and the power to commumcate what they are. Now if, as Aristotle

S, BROADIE

402

:i

'j.

· I . :. ,

�

believes non-human natures act unawares for the sake of ends, and regular! achieve those end� by natural strategies of astounding ingenuity, then what purpose is served by reflection? It cannot be m general that reflection is necessary or even useful for accomphshmg ends, since performance is faultless in countless cases where refle ct1on never _ enters. On the other hand, it is natural for human bemgs to act reflectively, and nature does nothing in vain. We must therefore suppose that human beings are distinguished by havmg ends_ that can _ only be achieved through reflection, and through the givmg and takmg of reasons. One such end that concerns us here is the transmission of forms of activity that are saved from dying with the individual on ly by _ the being taught. Refleclive rationality, according to this approach, is _ means by which human nature compensates for its own failure to provide genetically all that it needs in order to flourish and . continue. Non-human natures, on the other hand, are genetically adequate for their own needs, including the need to reproduc� in . kind. In comparing such natures to craft, Aristotle prescmds from JUSt the feature of craft that so impressed Socrates and Plato: the craftsman's command of the reasons for doing as he does. (17) The crafts develop; and even when a craft is well worked out, further intelligence is often needed to apply it effectively. There is room for thoughtfulness, inventiveness, and an exploratory attitude to facts and objectives. With nature, by contrast, it is as if the objective has never not been finally formulated and the means completely known. Thus when Aristotle likens nature to the craftsman, he has in mind the craftsman as already effeclively in aclion, and leaves out of sight the processes, with all their doubtfulness, trial and error, by which someone sets himself up so as to be thus effective: whether by learnmg a skill, or by analysing the objective, or by taking closer stock o f the _ _ is facts. It is perhaps at this stage that distinctively mental activity most apparent. We wonder and deliberate, not yet being ready to act _ m with the assurance of those whose knowledge is totally immanent their action. The thinking in action of the utterly accomplished performer is not alongside, above, beyond or even about his performan­ _ _ ce: but is that performance. It is the exercise of achieved craft that Aristotle has in mind when he compares craft with nature. The danger of the analogy is not that it will psychologise the conc�pt of nature , but _ that it will de-psychologise the concept of craft.17 For there is no . distinguishable psychological dimension to the self-contamed perfect10n of craft that figures in the analogy. This ideal efficiency essentially (1f

(17) I-lowever, this remark should be balanced by the observation that in the Ethics Aristotle treats craft as a paradigm for practical reflection. It is only in the Physics (see note 12} that he says that craft does not deliberate.

I I

J

NATURE/ AND CRAFT IN ARISTOTELIAN TELEOLOGY

403

it is the efficiency of craf() rests on a great deal that is not itself; but to acknowledge these other aspects would be to destroy the analogy. (18) These considerations may set us wondering whether the analogy's true direction is not the reverse of what it seems. If, as I have suggested, the notion of craft presented here is a false abstraction, then we are not dealing with a familiar concept of craft at all. Nor is it easier to understand than that for which it is supposed to be the model. Even the term "craft" is a misnomer, since the true referent here is not human craft as we actually find it, but end-directed automation. What more telling example of this could Aristotle show than the natural activity of organisms? Taken in its usual sense, the craft-analogy fails; but perhaps this argument shows that it was not needed in the first place. Could we even make sense of what passes for craft in the context of the analogy if we did not already grasp the idea of a type of teleological functioning that needs no roots in reflection?

I I

B1oux;IE, LOOIOUE l!T MtrfJ'HYSIOUE

!

chez

Alusrora

8eminaite CNRS·N.S.F., 1987

Editions du CNRS, Paris, 1990.

DE LA PHRONESIS ANIMALE JEAN-Louis LABARRI E RE

Aristote qualifie souvent les animaux, ou du moins un certain nombre d'entre eux, de phronimoi, mais il ne definit jamais clairement ce qu'il entend par Iii. On s'accorde generalement pour dire, en premiere approximation, qu'il designe par phronesis un certain type de connais­ sance et un certain type de comportement qui, tous deux, depassent Jes produits de la simple sensation. Le celebre debut de la Melaphysique ferait ainsi echo a ce passage de l'Ethique a Nicomaque OU Aristote mentionne que « certaines des betes sont dites prudentes (ton lherion enia phronima phasin einai) » parce qu'«il apparalt qu'elles possedent une faculte de prevoyance a propos de leur propre vie (peri Ion aulon bion echonla phainelai dynamin pronoetiken) » (VI , 7, 1 14 l a26-28). Ma question sera simple : quelle est la nature de ce qui « apparalt» ainsi et comment cela apparalt-il ? Quel type de connaissance et quel type de comportement une telle phronesis implique-t-elle par difference d'avec Jes animaux qui en seraient prives ? Quel rapport cette phronesis entretient-elle avec celle des humains ? Je conduirai cette analyse en deux temps : Tout d'abord, je m'efforcerai de construire un « ideal-type » de l a phronesis animale afin d e chercher a e n eclairer l a nature - question que, me semble-t-il, nous ne pouvons eviter, meme si Aristote n'en traite pas ex professo - , puis je tenterai de verifier la pertinence de cette construction en analysant Jes comportements animaux qu'Aristote qualifie explicitement d'intelligents et/ou prudents. Deux ensembles de remarques s'imposent toutefois prealablement afin de nous garder de toute interpretation globalisante. Le premier sera relatif au lexique utilise par Aristote dans Jes traites zoologiques. En effet, sa souplesse, mais aussi sa richesse, interdisent de le rigidifier, d'oU !'impression d'un vocabulaire fluctuant, apparemment peu rigoureux, impression renforcee, nous le verrons, par un frequent usage de termes traditionnels po � r designer des qualites ou defauts conventionnellement

L. LABARRIERE

DE LA PHRONESIS ANIMALE

done tout d'abord qu'en ces attri bues a tel OU tel anim al. Remarquons ment, pour ne P,as dir_e !erem indif r traites le Stagirite semble employe_ sis pour designer « l mtelh­ synonymement, Jes termes phronesis et syne u'il s'agit des hum ams, 11 gence prati que » des anim aux1 . Or, lorsq rativ e», de la synesis distingue Ja phronesis dite epilaklike, « impe st-il done advenu de cette « seulement critiq ue», krilike mono n2. Qu'e uise mem e des lors _ que distinction pour les anim aux ? La question s'aig ote !ait de la phronesis la !'on se souvient que dans la Polilique, Arist au gouverne, le premier qualite propre au gouvernant (archonlos} et non menos) de la 110.te et non au etant comparable au 110.tiste qui se sert (chro sans doute vain , au nom luthier, comp arab le, lui, au gouverne3• S'il est rce � de retro uver de s'effo de d'un e rigueur a mon sens ici hors de propos, mo1n s, tel sera mon pas reste telles distinctions chez les anim aux, il n'en hum aine son� , sous nesis phro la de propos, que certains des caracteres les aux an1m aux ferab trans , suite la des moda lites que je preciserai par e Jes textes ne mem bien d quan i, doue s de phronesis-synesis. Ains phronesis de la synesis ' leur permettent pas de distinguer a leur propos la . a un quelc onqu e pouvo1r de phronesis ou synesis renvoie sans doute dit de penser que chez eux commandement et ii n'est peut-etre pas inter eulement critiq ue»- _ N ean: I' aislhesis puiss e jouer ce role de !aculte « � peut fort b1en contmuer a mains, malgre un tel usag e, le terme phronesis ree des anim aux reputes ne designer que la « prud ence » un peu timo du Jievre (lagos, dasu pous }, !aches (deilos}, ainsi du cerf (elaphos) et nt a !'extreme 11uidite de leur sang caractere dO. physiologiquement auta . qu'il la grosseur de leur cceur 4• x s01ent Jes passages du Remarquons ensuite que bien que nom breu nous , ou bien de logismon et corpus privant les anim aux de dianoia, de lanl que lei, nom b reux sont en ent bien si\r de logos •, soit de raisonnem _ oyant, que ce s01t d1rrcte­ empl s gique egalement ceux des traites zoolo ees, le terme diano.ia et ment ou indirectement dans des expressions deriv ou intelligence anim ale'. Or, meme celui de nous pour designer la pensee ent brutalement le reste du si !'on veut que ces empl ois ni ne contredis ment metaphor�que, alors 1� corp us, ni n'aient un sens trap vague ou pure ne par la quelq ue chose qm !aut bien se resoudre a ce que Aristote desig

dans la pensee animale permette de parler d'une phronesis en un sens suffis� mment fort, quitte a relever par la suite son anthropomorphisme. En d autres termes, 11 faut admettre, me semble-t-il, que Aristote vise par 13 quelque chose comme un « raisonnement». Certes ce n 'est ni le logismon de cette partie de l'ame logique qu'il nomme lo logislikon, « �alcul �tr1ce »7, et e� core m ?ins le syllogismos scientifique, mais c'est 1 _ b1en, s 11 faut en cro1re ce qm est d1t de la phanlasia dans le De Anima, I I I, 10, 433a !0-14, « quelque chose comme une pensee » (Ms noesin tina), laquelle sorte de pensee est « pratique» (ho praklikos) puisqu'elle « calcule en vue d'une fi� » (nous de ho heneka lou logizomenos) . Etant donne que cela vaut auss1 et surtout pour les animaux8, prives dans ce meme passage de noesis et de logismos, tout tient done a la valeur que nous attachons a ce Ms. Nous le retrouverons lorsqu'a propos des dauphins fongant vers la surface, A�istote ecri t qu'ils retiennent leur scuffle hOsper _ _ analogisamenol « comme s Ils calcula1ent » la distance ou le temps qui Ieur reste (H.A . , IX, 48, 63l a27). Le probleme sera le meme lorsqu'il s'agira des r nombreuses 1m1tat10ns » (polla mimemala) de la vie humaine chez les petits oiseaux qui montrent ainsi « la precision de leur raisonnement» (ten frs dianoias akribeian, H.A., IX, 7, 612bl8-21). Plus generalement, ams1 que nous le verrons infra avec l'H.A . , V I I I , I , c'est tout le probleme de l'analogie. Ma derniere remarque relative au vocabulaire d'Aristote portera tout autant sur sa richesse que sur son caracter!3 souvent traditionnel : panourgia, eumkkhanos, eubiolos, ergalikos, lekhnikos, soot autant de termes frequ��ment employes par Aristote9 et que, d'Homere a Oppien, _ les Grecs ut1hserent pour qualifier des comportements intelligents. Plus partic� Iiere �ent, ils s'e n servi r�nt pour designer cette sorte d 'intelligen­ . , et dont nous devons a la sagacite de melts ce rusee qu tls nomma1ent

406

J,

61 lal6 ; 6, 612a3, 612b l ; P.A. , II, 2, (1) Cf. pour Ja phronesis, p. ex. H.A. , IX, 5, is : H.A. , VIII , I , 588a24, 29, 589al ; IX, 648a6, 8, 1 0 ; G.A., III, 2, 753a ll-12 ... Pour synes 46, 630b21 ... (2) Cf. E.N., VI, 1 1 , 1 153a8-IO. (3) Cl. Pol., Ill, 4, 1277b25-30. 4-30 ; Ill, 4, 667al0-21. (4) Cf. H.A. , I , I , 488b l5; P.A., 1 1 , 4, 650b l-14, 428a22-24, 429a6 ; 10, 433al l-12; 427b6 3, Ill, ; -Il 415a7 3, , 1 1 An., (5) Cl. De ... 3-5 332b 1 13, Vll, Meta., A, I , 980b27 et ss. ; Pol., 616b22, 30 ... Pour nous : ff.A. , {6) Cf. pour dianoia p. ex. H.A. , IX, 7, 612b20 ; 17, IX, 3, 610b22, voire P.A. , 1 1 , 2, 648a4-5.

407

{?) Cf. E.N..' VI, 2, 1 139a l l et ss. On se souviendra que cette <1 partie)> est aussi en VI, 13, I 144a l 4, ou bien encore to bouleutikon clans le De An. ' I I I ' nommee lo doxasitkon IO, 433b3 et 1 1 , 434a 12. �n rapprochant l� rech erc.he du desirable de la recherche du bien pratique, . (8) tient _ d'une explication unitaire du mouvement animal en effet par la le pr1nc1pe Aristote �et hu�� in) �t de !'action humai �e, �n m�me temps que le principe de leur difference grace a la distinction entre la phanlasta atslhi!ilke et la phantasia logisliki!-bouleulike operee dans le �e An., III, 1 1 . Cf. sur ce point D. J. Furley, Self Movers et J . B. Skemp, Orexis in De Anima !II 10, dans G. E. R. LL!oyd & G. E. L. Owen, Aristotle on mind and the senses Pr�ceed1�gs of the seuenth Symposium Arislolelicum, Cambridge, 1978; M . C. Nussbaum '. Arisiotle s IJ_e Mot� Animalium, Princeton, 1978; J .-L. Labarriere, Imagination humaine . . _ Phronesis, XXIX-N° 1 , 1984. an1male chez Ar1stote, et imagination nombreuses, en voici seulement quelques-unes tirees de tres sont rerences re Les (9) ,I H.A. Panourgta . : I, 1 , 488b20 ; IX, 8, 613b23 ; 37, 621 b29. Eumechanos et eubiolos : IX, 736, 614b34, 615a18, 616bl0, 1 1 , 13, 20, 27, 34, 620a21 ... Tekhnikos : IX, I I , 615a19 ; 13, 616a4 ; 34, 619b23 ; 36, 620bl0 ... Ergatikos, ergasia, ergazonlai : IX, 38-43, 622b19, 24, 26 ; 623b26, 624b31, 627a6, 9 . . Notons encore qu'une des trois sortes d'araignees est dite sophOlaion, 623a8.

408

J.

L. LABARRIERE

M . Detienne et J .-P. Vernant de nous avoir montre les multiples tours". Reste a savoir comment apprecier ! 'usage qu'en fait le Stagirite. Certes, doter Ies oiseaux d'une intelligence fine et « familiale» n'avait nen de bien original ainsi qu 'en temoignent ces vers de I' Elecire de Sophocle :
Parmi les oiseaux du ciel, quand nous voyons les plus inte.lligents

(phronimOlalous) prendre so in d � nourrir ce17x dont ils ont reQU la vie et le�

bienfaits qui soutiennent la vie, pourquo1 nous les huma1ns, pourquo1 d'ingratitude payons-nous nos parents ? » (1058-1061). . ,,1,• � j'

Mais faut-il pour autant supposer qu'Aristote ne se degage de cette ' tradition que fort difficilement, meme lorsqu'il s'y oppose, et qu'il paye ainsi sa dette au folklore ainsi que tend a le soutenir G. E . R. Lloyd ? n Des !ors Ies traites zoologiques d u Philosophe, lorsqu'ils dotent les ' animaux de phronesis ou gratifient tel ou tel animal de tel ou tel caractere seraient en contradiction avec les forts enonces de l'Ethique a Nicomaq�e selon lesquels il ne s3.urait y avoir d'animal vertueux12• Mais, durcir ainsi le trait, serait oublier, J. G. Lennox l'a fail remarquer13, qu'il n'est pas interdit d'en parler metaphoriq �ement, c'est-a-dire e� un sens mains strict, ce que fait precisement Ar1stote dans sa z�olog1e. ,. Or c'est bien cet usage qui m'interesse ici. Autrement d1t, plutot que de me demander ce qu'il en est de la teneur des propos d'A�istote en les rapportant a ceux de ses predecesseurs ou aux « on d �t » qu1 �va1ent cours et en comparant ses intentions affichees avec le tr1but qu 11 paye au folklore et a l'ideologie, je me demanderai, sans pour autant meconnaitre le contexte mais en le mettant entre parentheses, ce qu'il en est dans Ies textes memes du Stagirite. Plus precisement, ma question sera celle-ci : quand bien meme Aristote payerait un cert�in tribut a la tradition, quand bien meme Ies hvres V I I I et IX de I H.A. sera1ent marques par un net anthropocentrisme et anthropomorph1sme qu1 a pu faire douter de leur authenticite14, il n'en restera1t pas mo1�s que

(10) Cf. Les ruses de /'intelligence. La metis des Grecs, Flammarion, Paris, 1974. (11) Cf. Science, Floklore and Ideology, Cambridge, 1983, p. 18-43. (12) Cf. E.N., III, 4, l l l lb6 et ss (ou E.E., II, 10) ; Vil, 5, 1 1 47b3-5 ; Vil, 7, 1 149b31-1 150a I . (13) Cf. Demarcating Ancient Science. A discussion of G . E . R : Lloyd ; Science, Folklore and Ideology, clans J. Annas, Oxford Studies in Ancient Philosophy, vol. III, Oxford, 1985, p. 307-324. (14) P. Pellegrin, alors meme qu'il etudie Jes no�ions d � ge�r.e,. d'es�ece, e: d � . difference specifique, a Jui aussi insiste sur l' �nthr?pocentn�me anstotehc �en qut se revele a travers sa doctrine de l'analogie, cf. La class1ficallon des anu �aux chez Ar1�tote, Pa�·.�· 1982, p. 73-137 et plus particulierement p. 103-1 14. En ce qut concerne I authenllc1te_ des Jivres VIII et IX de l'H.A., malgre le recent article de P. Huby tenant pour leur inauthenticite et s'effor�ant de montrer ce qu ' ils doivent a Theophraste, cf. Theophrastus in the Aristotelician Corpus, with particular reference to Biological Problems, dans A. Gotthelf, Aristotle on Nature and Living Things, Pittsburgh-Bristol, 1 9�5, avec G. E. R. Lloyd, op. cil. p. 21, je me rallie aux arguments particuliere�ent con:a � n�uants de D. M. Balme, Aristotle's Historia Animalium : date and autorsh1p, 1981 (1ned1t).

T DE LA PHRONESIS ANIMALE

409

l'interprete doit s' interroger sur la coherence des differents enonces au sein du corpus et ne pas precipiter les reponses. Ainsi par exemple, malgre le caractere traditionnel d'un certain nombre de termes pour designer !'intelligence animale, rien n' assure a priori qu'ils renvoient a une meme qualite egalement ou plus ou mains partagee par un certain nombre d'animaux. Peut-etre richesse et souplesse ont-elles pour fin de signifier qu'il serait preferable de parler de phroneseis animales. Pour parler comme Descartes, gardons-nous done de toute praecipilalio et de toute praevenlio. M a deuxieme grande remarque preliminaire sera une nouvelle fois r.elative aux expressions memes dont use Aristote. En effet, lorsqu 'Aris­ tote mentionne sans plus s'y arreter }' existence d'une phronesis, c'est bien souvent au sein d'une echelle et ii emploie alors des comparatifs. C'est le cas des fameuses premieres lignes de la Metaphysique ou Ies animaux doues de memoire sont dits « plus intelligents » (phronim6tera) que ceux doues seulement de sensation. Certains seront d'ailleurs non seulement plus intelligents mais encore « plus doues pour apprendre » (malhelikotera) s'il sont de surcroit doues d ' ouie, ce qui n'est pas le cas de tous Jes animaux doues de memoire, ainsi de I' abeille et des autres animaux du genre15. C'est aussi le cas dans l'Hisloire des Animaux, V I I I , I , 589a l-2, ou, concluant un passage consacre a l a reproduction et a l'elevage des petits, Aristote ecrit que vivent d'une fagon plus politique (polilikoteron) avec leurs petits, ceux qui sont doues de memoire et sont « plus intelligents » (synetotera). C'est encore plus evident dans un passage parallele de la Generation des A nimaux ou en I I I , 2, 753a7-17, le Stagirite etablit entre Jes animaux une gradation relative aux sentiments et aux soins qu'ils portent a leurs petits : seuls les « plus intelligents » (phronimotera) s'interessent a leurs petits jusqu'a ce qu'ils soient eleves, et ceux qui «participent le plus a !'intelligence » (malisla koinonousi phroneseos) conservent des habitudes en commun et de l'amitie (synetheia kai philia) avec leurs petits une fois cet elevage termine. C'est le cas des hommes et de certains quadrupedes, mais non des oiseauxie. La phronesis animale semble done susceptible de degres et, s'il y a des animaux plus intelligents que d ' autres, cela doit signifier que les animaux inferieurs ne sont pas totalement denues d'intelligence mais qu'ils en ant mains - au demeurant, l'aislhesis n' est-elle pas deja une

(15) One f8.cheuse habitude assimile souvent les animaux mathetika et Jes animaux n'en est rien : si Jes hemera sont des mathetika, tel n'est pas le cas de tous les ainsi que le souligne l'H.A., IX, 1 , 608a17-21, qui attribue la mathesis et la aux animaux certes doues d'ouie, mais en soutenant qu'ils « apprennent1> aussi bien des hommes que de leurs semblables. (16) On trouve des formules voisines pour designer \'intelligence des biches : he elaphos ouch hekista dokei einai phronimon, H.A., IX, 61 la15-16. De son cote, !'elephant a une intelligence « h yperbolique », 46, 630b21 . hemera. II malhetika, didaskalia

i

l

'

i '

Ii :

)\ I

410

J.

L.

LABARRIERE

dynamis krilike ? A l'inverse, parmi ces animaux plus intelligents, certains le sont encore plus que d'autres, et leurs capacites sont perfectibles grace a l'ouie. Ajoutons encore que la phronesis ou synesis animate a affaire, meme faiblement, a }'experience et a l'habitude (Meta. , A, 1 , 980b27 ; Pol., V I I , 13, 1332b4), ce que souligne, dans l'H.A., V I I I et IX, aussi bien la richesse du vocabulaire dont se sert Aristote pour designer !'intelligence des animaux que !' attention qu'il porte aux sentiments « communautaires » des animaux. Or, s'il existe un lien entre le developpement de l 'inte!figence et celui des habitudes familiales, Jes choses deviennent nettement plus err1brouillees car les diverses hierarchies ne se superposent pas aisement.. Ainsi, les quadrupedes qui, avec les hommes, sont les animaux les plus intelligents 17, ne sont pas necessairement (bien qu'un certain nombre d'entre eux s'apprivoisent facilement) Jes animaux qui ont l'ouie la plus fine et la voix la plus deliee : ils sont sur ce point nettement depasses par Jes oiseaux (P.A., I I , 17, 660a l4-bl 1). Or Jes oiseaux passent pour s'attacher mains longtemps a leurs petits et etre mains intelligents que Jes quadrupedes, bien que ce soit chez Jes plus petits d'entre eux que !'on trouve, nous l'avons deja vu, de « nombreuses imitations » (polla mimCmala) de la vie humaine ainsi que « la sfi.rete du raisonnement» (ten tes dianoias akribeian). I I y a done bien la « des choses a rechercher» et ii faudra user, pour parler une nouvelle fois comme Descartes, de circumspeclio.

I. Logikos Qu'en est-ii done de la nature et du statut de cette intelligence animale? D'emblee nous nous trouvons confrontes a un epineux probleme puisque, aussi surprenant que cela puisse paraitre, ii semble bien que !'intelligence animale entretienne avec celle des humains une double relation : celle d'une ressemblance modulee selon l'ordre du plus au mains et celle, bien connue, de l'analogie. En effet, l'H.A., V I I I , l , 588a l6-b3, situe l a synesis a ces deux niveaux d'analyse : II y a, en effet, chez la plupart des autres animaux, des traces des etats de 1'3.me (ichne ton peri ten psychen tropOn) qui, chez les hommes, presentent des differences plus manifestes (phanerOieras las diaphoras). En effet, docilite et sauvagerie, douceur et rudesse, courage et 13.chete, crainte et temerite, emportement et fourberie, et (une ressemblance) de !'intelligence a !'egard du raisonnement (tes peri ten dianoian syneseOs) sont des ressemblances (homoioleles) avec l'homme qui existe chez bea_ucoup d'animaux, comme ce dont nous avons parle a propos des parties

<1

(17) Nous verrons par la suite que cela vaut aussi des vivipares tels les dauphins ou Jes phoques.

DE LA PHRONESIS ANIMALE

411

org� niques. Car certains (etats) (la men gar) different selon le plus et le mains (tOi_ mallon kai hetton diapherei) a l'egard de l'homme et il en va de men;e pour l' �omme a l'e&ard de beaucoup d'animaux (en e ifet certains de A ces et�ts (en �a gar ton loioulOn) existent plus chez l'homme, d'autres au contr.a1 �e (enza de) p lus ch:z le�. autres ani� aux), d'autres au contraire (la �e) different selon I analo_g1e (t_oi analogon dzapherei) : comme .il existe chez l h_omm � art, sagess� et 1ntelhgence (lekhne kai sophia kai synesis), ainsi . quelque autre faculte naturelle de meme ex1ste-t�1l chez cer�a1ns an1maux sorte (its hefera fowute physifre dynamis)- '> (trad. J.-L. L.)

S'il se�ble legitime de privilegier la relation analogique, reposant sur une difference de nature et non de degre puisque Jes animaux sont par definition �rive� du logisti�on humai � , il n'en reste toutefois pas _ que « l . 1ntelhgence relative au ra1sonnement» fait partie des mo1ns ressemblances susceptibles de degre. C'est bien la ce qui est surprenant pour qu1 ne retiendrait que l'existence d'une certaine faculte naturelle permettant a certains animaux de presenter un analogon de tekhne de sophia et de synesis, mais c'est pourtant ce qu'ecrit aussi Aristote da �s la parenthese des lignes 27-28 : le balancement qu'on y observe en enia gar . . . enia de renv�i � bien aux « traces des tropes de l'ame » des Iignes 1920. Or ces « quahtes psycholog1ques », qui comprennent l'enigmatique . ex�ress1on au gen1tif tes peri ten dianoian syneseOs, sont, dit alors Ar1stote, des ressemblances susce ptibles, selon Jes cas, de differer en degre . Ce balancement en enia des hommes aux an1maux et 1nversement. gar . . . enia de me semble rendre done tres delicat toute interpretation du balancen_ient general en la men ... ta de (Iignes 588a25-31) qui reserverait la prem1ere branche aux « caracteres », alors seuls susceptibles de degre, et la seconde aux « facultes intellectuelles » susceptibles seulement ' d'analogie. Au demeurant � es lors qu'il s'agit d'animaux1 exclure ce qui ; concerne les facultes mtellectuelles de ce qui concerne Jes caracteres (ethe}, ne semble pas correspondre a la demarche du Stagirite Jui-meme. Rema;quon� en � f�et, avant meme d'en venir a cet analogue de faculte, _ 1c1 operee par Aristote - aussi difficile soit-elle a que I mclus10n interpreter - n'est nullement unique. Je n'en prendrai que deux exemples. Dans l'H.A., I, 1 , 488bl2-28, lorsqu'il enumere et hierarchise <des differe� c�s relatives au caractere » (diaphorais kata to ethos), I . 12), . Ar1stote n hes1te pas a dire que seul l'homme est deliberateur (bouleutikon monon, I. 24) tandis que d 'autres animaux que Jui sont aussi . et de la capacite d'apprendre (mnemes kai didakes doues de memo1re I . 25). Et ii conclut ce passage en ecrivant qu'il reviendra sur cette etud � des c�:acteres et des manieres de vivre (peri ta ethe kai tous bious, I. 27), ce qu 11 aborde dans Jes hvres V I I I et IX d'ou sera tire mon deuxieme exemple. Dans les premieres lignes du livre IX, Aristote compte en effet la phronesis, au meme titre que le courage OU }a douceur, parmi les ethe ou Jes pathemala possedes par Jes animaux en raison de I' existence en eux d'une « certaine_ faculte naturelle » (tina dynamin physiken, 608a l 3-l 7).

412

DE LA PHRONESIS ANIMALE

J. L. LABARRIERE

alors Nous voici done ramenes a cette <(facult e nature lle». Toutefois, chez nce !'existe de vue qu'en ee postul it sembla qu'en V I I I , 1 , elle ne est elle , synesis de et sophia de , lekhne de on analog d'un certains anima ux des ent dite en I X 1 exister chez les anima ux afin qu'ils possed ion soit affe �tions co �re�ponda ntes a celles deS humai ns. Que �on extens venons nous e alors plus grande ne fait qu'aviver notre p�obleme pmsq � _ III, 1, V H.A., I de de voir qu'une lecture attentive des premieres hgnes de plan le sur synesis a l a tendait a montrer que quelque chose de relatif des degre en r differe i>, la dianoia pouvait, comme d 'autres « tropes ent de ces hommes aux anima ux. Le chevauchement et l'entrecroisem bien quand , temps r premie un en textes me semblent done montrer ent uemm conseq et giee, privile etre t meme la relation analogique devrai l� anima s synesi ou sis phrone la de statut la difference de nature, que le a parter en puisse !'on que bien semble 'il ne va nullem ent de soi puisqu . fagons deux niveaux ou de deux faculte Si maintenant nous nous tournons de nouve au vers cette de de la Le ? est-ii qu'en , l , I I I V en tee presen est qu'elle naturelle telle es certam que 588a28 repondant au la men de la ligne 25, introduit done des celles vec a qualites n'ont chez les anima ux qu'un rapport d'analog1e . en donner un humai ns. Aristote ne Ies enumere pas, ma1s 11 se contente d : ii existe 30) l. z66n, liln exemple valant pour certains animaux (eniois ce qm que sorte meme de lle chez ceux-la quelque autre faculte nature . synesis de et sophia de , tekhne chez I'homm e Jui permet d'etre doue de en faculte cette de tion ermina Remarquons tout d 'abord qu'a !'indet . arnma ux correspond une autre de nature differente : celle relative aux passage ce tout en s01t ne l arnma un qu'auc doues de cette facult e. Bien certain un a if attent e qu'etr peut ne on explicitement mentionne, de la et s blance ressem des vant aupara dit etait decalage puisque ce qui e no.mbr difference du plus au moins sembla it valoir pour un grand Jes pour e, d'anim aux (pros po/la, 1 . 26), OU, par opposition a l'homm s1 l'on vo1re, 28), I. zOois, allois iois (en l genera en ux autres anima i? is (en ux remonte aux « traces 1>, pour la plupart des autres anima s cerlain de nan � pleislois kai ton allon zoon, 1 . 19), or il, s'agit mainte valeur une avo1r 1t anima ux (eniois, 1. 30). Certes, cet mdefirn pourra plus voisin e de celle des deux enia de la ligne 27, mais ii semble tous : tion restric » e certain « raison nable d'adm ettre qu'il introd uit une bien Ies animau x ne possedent sans doute pas une telle faculte tre utpe t so ne t stemen manife ent � developpee, et ceux qui la possed _il n y a meme b1en quand is, pleislo un pas legion. De la un eniois et non re des peut-etre la qu'un effet de la difficu lte a connaitre le caracte . l-13) 608al , 1 IX, , . (H.A animau x difficiles a etudier et, Reste maintenant a preciser l a nature de cette faculte analogue S1, partant ' ! 'intelligence pratique des animau x qui en sont doues. l4486a comme y invite Aristote, nous nous tournons vers l'H.A., I, 1,

�

413

b24, oil ii explicite les differentes sortes de ressemblances entre Jes parties du corps, nous devrons nous souvenir qu 'ii en existe trois sortes : soit les parties se ressemblent par leur forme a l'interieur d'une meme espece (un nez ou un ooil d 'homme a un autre nez ou autre reil d'homme) ; soit elles se ressemblent en different par exces ou defaut a l'interieur d'un meme genre (certains oiseaux ont le bee long, d'autres court) ; soit enfin elles ne se ressemblent ni specifiquement, ni selon l'exces et le defaut, mais par analogie (l'arete est !'analogue de l'os). II existe done chez certains animaux une faculte naturelle jouant la meme fonction que !'art, la sagesse et !'intelligence chez l'homme, mais, bien que naturelle, elle ne saurait etre de meme nature que ce qui joue ce role chez l'homme, ni par consequent differer selon le plus ou le moins avec celle de ce dernier. Voila qui devrait etre clair si, d'une part, nous n'avions deja vu que ce qui a trait a la synesis sur le plan de la dianoia pouvait etre classe parmi les caracteres semblables susceptibles de differer en degre, et si, d 'autre part, cette faculte naturelle ne s'appliquait qu'aux facultes intellectuelles, ce qui, d 'apres l'H.A,, I X , 1 , 608a l3-17, n'est pas. D'un autre cote, les chases ne sont pas vraiment plus claires si l'on se penche vers ces caracteres differant selon le plus et le moins. En effet, d'apres l'H.A., I , 1 , 486a21 -bl7, ce qui differe selon l'exces et le defaut (le plus et le moins pouvant etre consideres comme tels) differe a I1interieur d'un meme genos. Comment done, en toute rigueur, une telle definition pourrait-elle s'appliquer a la comparaison des caracteres entre les humains et tel ou tel genre animal ? Nous pouvons sans doute deduire de la tout autant le privilege souvent accorde a l'analogie que les critiques adressees a Aristote pour son manque de rigueur en zoologie. Cependant le terme genos etant susceptible d'une plus ou moins grande extension, rien n'interdit de continuer a parler assez rigoureusement si, pour que la difference selon le plus et le moins reste a l'interieur d'un meme genre, on entend par genos ce qui designe le genre animal, doue de sensibilite, par opposition au genre vegetal depourvu de sensibilite. Cela, a condition de sous-entendre que l'on parle alors des caracteres en un sens moins rigoureux que dans les Elhiques puisque !'on sait qu'Aristote refuse a l'aisthesis d'etre au principe d'une quelconque action en tant que telle parce qu'elle est commune aux humains et aux animaux (E.N., I , 6, 1098a2-3 ; VI, 2, 1 139al8-20). Reconnaissons toutefois que la rigueur ainsi donnee a la difference selon le plus et le moins ni n'eclaire ce qu'il en est de cette faculte (( analogue», ni n'indique quels pourraient etre les animaux a qui elle appartient au premier chef. C'est pourquoi, afin de mettre en rapport ces deux sortes de ressemblances et afin de preciser de quels animaux ii s'agit, nous pourrions supposer, puisque ce qui est dit des « traces des qualites psychologiques » doit avoir des animaux pour support, que le

414

J.

L . LABARRIERE

genos au sein duquel s'exerce la difference selon le plus et le moins,

renverrait non au genre animal en general mais au genre des sangu� ns, tandis que ce qui est dit de I' analogon vaudrait pour les non-sangums. Voila qui expliquerait que nous pourrions trouver la synesis des deux cotes de I' alternative ouverte par le balancement en la men ... la de. Nous pourrions des !ors comprendre ce passage parallelement a celui des P.A ., . I I 2 648a2- l l , qui rapporte les qualites de la sens1b1hte ou de I'i�te lligence a celles du sang ou de ce qui en tient lieu : . « Un sang plus epais et plus chaud donne plus de �o��e! tandis, � u'ul?- sang

plus Ieger et plus froid donne plus de . s� ns1b1hte et d 1ntelhgence (aisth€tik6leron kai no€r6leron). La meme d1ffere� ce se r� trouve clans les humeurs qui correspondant au sang. C'est pour� uo1 l�s abe1lles et les autres . _ . animaux de cette espece sont doues de . plus d 1n.telhgence (phronimofera) que bien des animaux sanguins et, parm1 ces dern1ers, ceux dont le sang est froid et Ieger sont plus intellig'ents (phronim6lera( qu.e ceux dont le sa�g a Ies qualites contraires. Mais l� � ieu:-: est encore � a� o1r le � ang chaud, leger et clair : car les animaux a1ns1 fa1ts sont doues a la fo1s de courage et d'intelligence. » (trad. P. Louis).

I

' ' I1 ,, '

i •. :

1 1. ,q,, , I\' L\1 1 •

1: 1· 1

c� L .i! : i

.•

: :·{1 '1it·1' · I I

!

Gi '

Pourtant aussi commode soit ce rapprochement, ii ne me semble pas vraiment' probant. En effet, � utre so.n ca �ac�ere eminemment conjectural (rien en ce passage de I H.A . � est d1t a propos de cette difference entre sanguins et non sangums), une telle hypothese entralnerait que seuls Ies non-sanguins seraient doues d'une telle faculte naturelle (( analogu e» a ce qui chez les hum �ins est art, s�gesse et intelligence, tandis que ceux qui se dif!erenciera1ent de ces dermers selon le plus et le mains, c'est-3.-dire les sanguins, n: aur�ient _que des traces de synesis. Voila q.ui s.emble fort improb able pm�qu 11 , ex1�te de nombreux animaux sangums mtelhgents, quand b1en meme I abe1lle ou la fourm1 depassent en intelligence les sanguins au sang epais (p. ex. Jes taureaux et Jes sangliers, P.A ., I I , 4, 650b33-65l a4). Remarquons en revanche que dans ce passage des P.A. les comparatifs .permettent . de mettre e� rapport la difference de nature (selon l 'analog1e) et la difference de degre (selon le plus et le moins) puisque des animau x possedant un analogue d � ceci ou de cela peuvent etre dits superieurs a certams de ceux qm different selon le plus et le moins. ,. On le voit done, Jes chases sont nettement plus complexes qu 11 ne pourrait sembler au premier abord. II semble bien que l'on puisse. parler de la phronesis OU de ]a synesis a deux niveaux d'anaJyse, celUI de Ja difference ou ressemblance selon le plus ou le morns et celm de la difference ou ressemblance selon l'analogie, niveaux entrecroises avec ceux des caracteres et des facultes. En effet, tandis que l'H.A., V I I I , 1 , 588a28-31 semble reserver a une physike dynamis indeterminee la fonction d� presenter chez certains animaux un analogue de ce qui chez l'homme est art, sagesse et intelligence, I' H.A ., IX, 1 , 608al3- l 7 sen,ibie de son cote ouvrir les fonctions de cette faculte naturelle pmsqu elle

DE LA PHRONESIS ANIMALE

415

permet alors aux animaux de presenter des palhemala et des hexeis correspondant a chacun (hechaslon) de ceux des humains. Les qualites ou defauts par lesquels humains et animaux different selon le plus et le mains, peu importe ici l'intensite de ces « traces i>, peuvent done etre egalement rapportes a cette faculte naturelle. De la la possibilite de generaliser l'analogie afin de ne pas parler seulement par metaphore ou seulement selon une tradition folklorique : c'est en fonction d'une certaine faculte naturelle, semble-t-il plus ou moins bien partagee chez les animaux, que l'on peut, a leur propos, parler de caracteres et de qualites intellectuelles. Mais a !'inverse, de la aussi, une fois cela entendu et si nous nous souvenons qu'hommes et animaux sont precisement des animaux et non des plantes, que l'on puisse en parler en terme de degre puisque nous nous situons alors au niveau d'une psychologie biologique et non au niveau d'une psychologie morale18. C'est en ce sens, me semble-t-il, qu'il faut comprendre !'inclusion de la synesis relative a la dianoia au sein des qualites pouvant differer en degre, inclusion plus nette encore, je viens de le rappeler, au debut de l'H.A . , IX, 1 . Toutefois, quand bien meme certains des effets de la phronesis animale peuvent, selon des modalites certes complexes, etre compares a ceux de la phronesis humaine, ii n'en reste pas mains que la phronesis animale ne saurait etre que fort differente de la phronesis humaine puisque les animaux, quels qu' ils soient, sont prives d'hypolepsis et de bouleusis. A strictement parler, je l'ai deja mentionne, la phronesis animale ne preside a aucune prdxis. Nous verrons cependant plus loin, nouvel exemple de la souplesse du vocabulaire aristotelicien, que le Philosophe n' hesite pas a parler de praxeis animales quand ii entend designer des « actes » ou « actions » depassant le strict mouvement local. II en va ainsi de ce qui a trait a la recherche de la nourriture, a leur conservation et a la protection de leur progeniture, autant d'actes qui manifestent souvent des comportements intelligents, voire « ruses ». On ne saurait done tenir la phronesis animale pour une simple metaphore, d'autant qu'un certain nombre d'animaux sont « plus prudentS >> que les humains, par exemple dans leur capacite a prevoir le temps qu'il fera (ainsi des abeilles, H.A ., IX, 40, 627bl0-13). C'est pourquoi ii me semble juste, en un premier temps, d 'en construire un « ideal-type» en transposant mulalis mulandis a la phronesis animale certains des caracteres de la phronesis humaine. En premier lieu !'intelligence animale ajoute a l'aislhesis son propre pouvoir de discrimination lie aux phanlasmala de la mneme et de la phanlasia. D'apres le De A n . , I I I , 3, 427b6-8 ii ne caracterise qu'un (18) Cf. W. W. Fortenbaugh, Aristotle : Animals, Emotion and Moral Virtue, Arethusa, 4, 1971, �· 137-165; Aristotle on Emotion, London, 1971, p. 26-30, 66-70.

416

J. L. LABARRIERE

certain nombre d'animaux et, nous l'avons vu, ii peut etre renforce par la possession de l'ouie. Autrement dit, elle implique des conduites qui ne sont pas seulement instinctives mais qui relevent d'une sorte de raisonnement. P. ex . , les dauphins lorsqu'ils poursuivent une pro"ie, retiennent leur sou!fle « comme s'ils calculaient» leur temps (hOsper analogisamenoi, H.A . , I X , 48, 63la27). De meme la seiche (sepia) passe pour « le plus ruse » (panourgolaton) des cephalopodes parce qu'elle !ache son encre pour se dissimuler et non par peur, tandis que_ le poulpe (polypous) est inintelligent (anoeton, H.A . ; I X , 37, 62Ib28-622a l4 ; P.A., IV, 5, 679a4-30). II semble en effet que nous puissions dire que de meme que la seiche r ntilise » (chretai, 62l b29 ; chreslhai, 679a9) son encre, de meme ce qui caracterise la phronesis animale c'est un certain usage des sensations et de l'habitude en vue d'un certain mieux-vivre et non du seul survivre. Le De Sensu, 1 , 436bl2-437al7, distingue ainsi deux fonctions, pour Ies animaux doues du mouvement local, des sens « externes », savoir l'odorat, la vue et l'ouie. D'une part, ils existent en vue de la conservation (s6leria) en guidant la recherche de !'utile et la fuite du nuisible, mais, d'autre part, « pour ceux qui sont doues de reflexion, ces sensations leur . appartiennent en vue de leur bien-etre (tois de kai phroneseOs lunchanousz lou eu heneka) : elles leur apprennent en effet a distinguer clans les objets beaucoup de differences (pollas ... diaphoras), d'oU leur vient la connaissance et des objets de pensee et des actions a faire (tOn noeton enginelai phronesis kai he lOn praklOn). De ces facultes, la _!l1eilleure, en elle-meme, c'est la vue, pol:lr Jes besoins necessaires de !'animal ; mais pour !'intelligence (pros de noun) et accidentellement, c'est l'oule . » (trad. R. Mugnier).

Levons d ' emblee une difficulte plus a pparente que reelle : bi en que la phronesis appartienne aussi a certains animaux, il pourrait sembler, en raison de la connaissance qu'elle procure ici, que ceux qui en sont·· doues sont seulement Jes humains. Une telle impression pourrait etre renforcee par ce que dit Aristote de I' ouie a la fin de ce passage : « l'ou'ie ne fournit que les differences de son (las lou psophou diaphoras, a La difference de la vue qui fournil des differences nombreuses et variees), et, po? r quelques etres aussi des differences de voix (las tes phones) ; mais, accidentellement, elle contribue a la pensee (phronesin) pour une tres grande part, car le langage est la cause de !'instruction (logos ailios esli tes mathese6s), non en lui-meme, mais indirectement : ii se compose de mots (onomatOn), et chacun des mots est un symbole (symbolon) . )> Cependant, malgre l'usage d'un vocabulaire caracterisant l'humain, ii n'en est rien puisque l ' H.A . , IX, I , 608al7-21 attribue a l'oule de certains animaux les memes pouvoirs : ils ont la capacite aussi bien d'apprendre que d 'enseigner (malheseos kai didaskalias) , parce que leur oule pergoit non seulement des differences entre les sons, mais encore entre Ies signes (ton semeion diaislhanelai las diaphoras). La double fonction de la vue, de l'oule et de l'odorat, ne car�ct�rise done pas

DE LA PHRONESIS ANIMALE

417

seulement les seuls humains, mais aussi les animaux doues de phroni!sis, quand bien meme c'est chez l'humain que cela est le plus evident. Qu'il y ait « recherche d 'un bien » pour certains animaux est encore confirme par le De An., I I I , 13, oil Aristote souligne que les sens autres que le toucher existent en vue du bien et non en vue de la seule vie (435bl9-25). Quelle est done la nature de ce lo eu ? S'il implique un certain type de connaissance et un certain type de comportement, qui tous deux ne se reduisent pas au seul instinct de conservation, nous pouvons encore remarquer qu 'ii semble impliquer aussi un certain type de vie communautaire, qu'elle soit gregaire, politique, ou seulement familiale, et l'ou!e (avec la voix) vient renforcer ce lien19• II y a lieu d' ailleurs de se demander si nous ne retrouverions pas ce lo eu dans le fait qu'Aristote se serve plus volontiers des termes eumechanos ou eubiolos que des termes mechanos ou biolos pour qualifier un certain nombre d'animaux, generalement des oiseaux. C'est ici le lieu de prendre au serieux cet enonce d' Aristote selon lequel on peut considerer la phanlasia aislhetike des animaux « comme une sorte de pensee . . . pratique parce qu'elle raisonne en vue d'une fin » (De An., I I I , 10, 433a9-21) lorsqu'elle s'unit au desir dans la recherche du desirable. II y a en effet deux traits de la phronesis humaine qui sont transposables a la phronesis animale : un certain rapport a !' experience et le caractere imperatif (epilaklike). Si la phronesis ou la synesis animale est superieure a l' aisthesis, voire a la mneme et a la phanlasia en ce qu'elle implique un analogue de raisonnement (une sorte de pensee pratique), c'est sans doute parce qu'elle implique un certain recul par rapport aux differences pergues par l'aislhesis et conservees dans la mneme. Cette « pensee)> est « imperative)> parce qu'elle commande en faisant usage de ce qui est propose. Autrement dit les conduites prudentes different des conduites seulement instinctives en ce qu'un quelque chose qui keleuei vient s'interposer. C'est bien semble-t-il le cas de la seiche lorsqu'elle !ache son encre : elle aussi a peur, mais elle se sert de cette peur et de son encre comme d'une arme. Sa ruse consiste a construire comme un rempart (phragma) autour d'elle en noircissant l'eau et en y revenant apres avoir fait mine de fuir. Plus generalement, les conduites « prudentes » demandent une certaine ma1trise de la temporalite afin d ' agir en fonction de !' experience tiree d'evenements passes, soit en fonction d'une certaine habitude. Reste a savoir commment le passe est utilise puisque si les animaux, ou du mains un certain nombre d'entre eux, sont doues de memoire, ils ne sont pas doues d'anamnesis, laquelle est proprement humaine car elle est

(19) Si l' Economique etait authentique, on pourrait encore se referer a I , 3, 1343hl 518, qui fait une place aux animaux domestiques et plus intelligents en les situant entre les betes brutes et le� humains car leur union est «plus articulee » (dii:rlhr6lai).

418

J.

L. LABARRIERE

bouleusis et zetesis. D'un autre cote, malgre leur capacite de prevoyance, ils ne pergoivent pas le futur comme les humains, puisque chez ces derniers il est objet de doxa (De Mem., 1 , 449b9-15)20• Au mieux, les animaux n'utiliseront que des « empreintes », mais cette utilisation releve-t-elle d'un rappel automatique ou bien d'un rappel quelque peu reflechi ? y aurait-il chez certains animaux participant faiblement a !'experience et a l'habitude, quelque chose comme une « proto­ reminiscence » afin qu'il y ait quelque chose comme un temp� d'arret ? Si Jes animaux ne participent que faiblement (mikron) a !'experien­ ce et a l'habitude, c1est bien entendu, que l'on se fonde sur Meta. , A, 1 , OU An. Post., I I , 19, parce qu'ils ne parviennent pas a unifier leur souvenir, a construire un logos. Jls sont done prives d'hypolepsis et ii est inutile d'y revenir. Mais si cette participation faible doit etre quelque chose et non un pur neant, alors cela signifie qu'ils ant un peu de ce qu1 constitue !'experience en tant que telle. Par consequent, meme bornee a ce que propose la mneme et la phanlasia, leur phrones is doit operer un . certain take-off sans l'operer totalement car alors les ammaux sort1raient du cadre trace par leur ilme sensitive. JI faut done sup poser, si le « peu » renvoie a quelque chose et non a rien, qu'il renvoie a quelque chose comme une sorte de « sous-proposition d'empiriel> parce que ces animaux-13. possedent quelque chose comme un sous-universel <( stabilise tout entier dans l'ame» (An . Post., I I , 19, 100a6-7). Autrement dit, en supposant qu'en 100a7 le « iou henos para la polla » n'ait pas son sens . platonicien mais vaille pour le « hen kala pollon » anstotehc1e� (I, 1 1 , pre-predicahf» 77a6), ces animaux doivent posseder un « sous-um verse! qui Ieur permet de ne pas agir a I' aveuglette et de ne pas la1sser leur sort a la chance. Eux non plus ne doivent pas tomber sous le coup du mot de Polos : meme faible, leur empeiria ne doit pas etre une apeiria (Meta., A, l , 98la4). Toutefois, c'est la faiblesse de leur empeiria qui leur interdit de developper une lechne. . Si Ieur simili experience ne leur permet pas de rapporter clairement des moyens a une fin, ii n'en reste pas moins, puisque ce son � des animaux superieurs, qu'ils ont en vue quelque chose comm� un « mte1:1x­ _ . au heu vivre » bien qu'ils ne prediquent pas a proprement parler. A1ns1, de la proposition d'empirie du style : « ce remede a gueri Callias, Socrate et quelques autres, done ii peut en guerir d 'autres», peut-etre ces animaux raisonnent-ils comme suit : « du frais, du vert, du flu1de, une biche 1 done une riviere pour etancher ma soif », cette deduction se faisant parce que la riviere a ete temporellement ass�ciee au frais, au vert (�es arbres), au fluide, a une biche en train de boire. Aussi , percevant certa1ns

·.

(20) On notera la curieuse description physiologique de ce propre humain : L'homme est pour ainsi dire le seul a ressentir des tressaillements (au cceur) parce qu'il est le seul a esperer et a pressentir l'avenir», P.A., I I I , 6, 669a18-20 (trad. P. Louis):

«

1 I

DE LA PHRONESIS ANIMALE

419

de ces elements a distance, !'animal entreprend de se deplacer en ne laissant done pas au hasard de ses promenandes le soin de lui presenter une riviere. II en irait de meme pour des « propositions» du style : « du rouge, un cri, du rugueux, m'ont permis de construire un nid». Tres certainement, si de telles propositions etaient formulees par un humain, ce serait, comme le dit Polos, de I1inexperience. Mais pour un animal, c'est sans doute de !' experience car, pour effectuer ce semblant de raisonnement, ii a fallu stabiliser sensations et traces sensorielles, soit marquer quelque chose comme un temps d' arret, c' est-a-dire faire un pas vers ce premier universel indifferencie (sianios iOn adiaphorOn henos, An. Post., 11, 19, 100al5-16) auquel les animaux n' accedent pas en tant que tel. En raison de la frequente comparaison entre les enfants et Jes animaux, ii n'est pas interdit de se demander si les animaux n' accederaient pas a quelque chose comme ce premier universel que les enfants manifestent en appelant tous les hommes « papa » (Phys., I , 1 , 184bl2-14). Ou bien encore n'en irait-il pas pour certains d' entre eux comme du passage de l'etat d'enfance a J'etat adulte puisque ce passage s' accompagne, dit Aristote, d'une stabilisation du desordonne (Phys. , V I I , 3, 247bl9-248a6) comparable iI ce qui s e passe lorsque !'on sort du sommeil ou de l' ivresse ? Je conviens volontiers que de tels rapproche­ m·ents soot hasardeux, mais enfin lorsque la biche manifeste sa phronesis en faisant ses petits sur des sentiers degages (H.A . , VI, 29, 578bl6-17 ; IX, 5, 61 lal 5-22), n'est-ce pas parce que tous Jes fourres lui paraissent dangereux etant donne qu'ils peuvent abriter d'autres betes tandis que les chemins sonl plus stirs puisque les betes fuient les hommes? Entre deux maux, les betes ou les hommes, la biche semble bien choisir le moindre, l'homme a ses yeux. JI y a done bien la quelque chose comme un raisonnement impliquant un certain acte de 1'3.me et ce raisonnement peut etre tres sO.r, comme chez les petits oiseaux. C'est pourquoi, si l'on se souvient qu' Aristote, dans la Meta. A , 1, souligne que parfois l'homme d' experience reussit mieux que l'homme de l'art bien que ce dernier l'emporte en sagesse, ii n'est pas interdit de transposer aussi aux animaux ce caractere de l'experience. De ce point de vue, ii faudrait meme aller jusqu'a dire que la phronesis animale puisse depasser en intensite celle de l'homme dans la conduite de certaines actions. On comprend ainsi que !'organisation de la ruche puisse passer pour une petite merveille. S'il est permis de pousser plus loin ces hypotheses, alors ii faudrait aIler jusqu 'a dire que ces capacites de « raisonnement pratique )> dans lesquelles la majeure du syllogisme serait une simili proposition d'empirie et non une simple « proposition d'epiihymia )>, sont autant congenilales qu'acquises. La stabilisation due a un acte de l'i\me serait le fruit de !' education et de !'experience, autrement dit de la coutume ou

420

j! :1 ' .i ', :\!

J , L. LABARRIERE

des habitudes. Toutefois, si un tel processus a lieu, ii reste en!erme dans le cadre de 1'3.me sensitive. Aussi, lorsqu'un animal exerce un acte de phronesis (c'est-a-dire lorsqu'il determine son comportement en fonction d'un lo eu d'apres un ethos acquis), cet acte, a titre d 'energeia, est bien Un acte intelligent (c'est-a-dire une position du bien), mais ii faut que ce bien soit « represente imaginativement» dans le cas particulier et telle est la fonction de la phantasia aisthetike qui, pour etre superieure a la simple sensation, y reste rattachee. Nous serions ainsi conduits a soutenir que pour que la phronesis animale ait un sens, ii faut qu'elle corresponde a un mouvement declenche par l'habitude et !'experience, et non par la seule sensation. Remarquons cependant avant de conclure qu'une telle reconstruction suppose que la memoire des animaux intelligents comporte quelque chose comme une proto-reminiscence qui ne soit ni bouleusis ni zeiesis. Bien que nous n'en trouvions trace, a ma connaissance, nulle part, il n'est pas interdit de se demander si l'un des traits de l'anamnesis humaine ne serait pas transposable aux animaux : le phenomene de co­ evocation des souvenirs peut etre d€:crit comme une sequence de mouvements (kineseis) s'enchainant en vertu d'une liaison associative (De Mem., 2, 45lbl0-16). Peut-etre y aurait-il alors chez certains animaux une suite mecanique d'empreintes sensorielles guid€:es par l 'habitude, mais Jes collections seraient plus hasardeuses que chez Jes humains car, sans doute, la contiguite temporelle l 'emporte-t-elle chez eux sur les liens de ressemblance, d'oU ces simili propositions d'empirie de type aberrant pour l'humain,· mais dans lesquelles une collection d'empreintes (du rouge, un cri, du rugeux) peut tenir lieu de « sujet» au meme titre que Callias, Socrate . . . Malgre son caractere eminemment conjectural, cette reconstruction n'en a pas mains, a man sens, sa coherence. Elle pousse jusqu'au bout ce fail que participer, meme faiblement, a !'experience et a l'habitude, doit signifier nettement plus que n'y pas participer du tout. Maintenant, est­ il possible de reconnaitre une telle phronesis dans Jes conduites animales, plus particulierement dans l'H.A., IX, livre le plus riche sur, ce point? Je dais avouer ici man. embarras et ma seconde partie sera dans une certaine mesure plus negative que la premiere. J'espere toutefois qu'elle aura pour elle d'etre plus certaine.

II. Physikos Renversant la demarche proposee precedemment, j e me demanderai si le modele heuristique propose n'est pas trop global. En effet, s'il peut etre satisfaisant pour !'esprit (eulogos), rien n'assure qu'il corresponde bien a la demarche aristotelicienne. Plutot que de chercher a construire

1 I

DE LA PHRONESIS ANIMALE

421

un « ideal-type » utile it l'interprete, n'est-il pas aussi pertinent de prendre acte du fait qu'Aristote ne s'en soil, semble-t-il, pas soucie ? Ne nous abuserions pas en parlant de la phronesis animale ? Devons-nous supposer qu'il n'existe qu'une seule et meme phronesis pour tous les animaux ? Autrement dit, la construction d'un tel ideal-type ne revient­ elle pas a faire comme si, a l'instar du genre humain, ii n'y avait qu'une seule espece animale ? Ne pecherions-nous pas en faisant
Si done !'on admet a partir de ce texte que la phronesis concerne la recherche du bien relatif it ses propres interets, it sa vie propre - ce qui justifie que l'on puisse parler d'animaux phronimoi et que ni Thales, ni Anaxagore, pour sages qu 'ils sont ne puissent etre consideres comme

(21 ) Les lignes 1 14la25-27 sont en effet controversees. Je n'entends pas entrer ici da ns Jes details �e. cette � polemiquel), car qu'on maintienne avec les MSS le aulo, ce que _ Bywater, su1v1 en France par Tricot, ou qu'avec Burnet on le supprime, le<;on suivie fa1t par Gauthier et Jolif, ii n'en reste pas moins que la phronesis differe de Ja sophia en ce qu'elle est variable et relative, ce qui est le point essentiel de l'argumenL La premiere le�ture, qu1. me semble preferable, rapporte cette relativite au bien poursuivi par l'animal _ que la seconde la rapporte a l'eleveur qui sait ce qui est bien pour chaque lu1�m�me, tand1s animal.

422

J. L. LABARRIERE

prudents - alors ii faut en conclure, quelle que soit la legon ici adoptee, qu 'ii y a di verses phroneseis, chacune etant relative au bien de chaque animal". Autrement
(22) On pourrait eventuetlement m'objecter que ce passage de l'E.N. porte sur la sophia et non sur la phroni!sis, par consequent on ne pourrait rien en conclure sur Jes phroneseis animales. Je ne crois pas cette objection recevable car, s'il est bien vrai que ce chapitre traite de la sophia, ii s'efforce de la distinguer de la phronesis, et te\ est le sens du passage sur lequel je m'appuie. L'argument (cf. supra n. 21) consiste a soutenir que la phronesis differe de la sophiO. en ce qu'elle est variable et relative, des !ors on ne saurait user du terme sophia pour designer un savoir variable et relatif sans se contredire, ce que font ceux qui nomment sophia !'art politique. En effet, ce dernier en tant qu'il releve de la connaissance de ses propres interets, est variable et doit etre proprement nomn1e phronesis (de 18 d'ailleurs que ce terme puisse egalement etre applique a la connaissance qu'ont Jes

animaux de leurs propres interets), or, si !'on nomme improprement cette connaissance sophia, alors ii y aura une multiplicite de sophiai, ce qui ne peut et doit etre en raison de l'unicite de la vraie sophia. On le voit done, ce texte sert fort bien mon argument relatif a la multiplicite des phroneseis, puisque ce que Aristote veut dire en disant qu'il n'y a pas une seule sophia s'appliquant au bien de tous Jes animaux mais une differente pour chaque espece, c'est que cette sophia-18 n'en est pas une, sauf a parler improprement, mais est, par definition, une phronesis. Notons encore que sur la variabilite et sur la relativite, on trouvera un argument du meme type a propos du plaisir dans \'E.N., X, 5, 1 176a3-9.

1 I

DE LA PHRONESIS ANIMALE

423

En laissant ici de cote les considerations relatives au climat2a, il me semble que c'est a partir de ces trois types d'« actes vitaux » (reproduc­ tion, elevage, nourriture) que l'on peut comprendre en quoi les animaux sont phronimoi. Quand bien meme en effet la phronesis animale renvoie a cette dynamis pronoetike relative a la recherche du bien pour sa propre vie, elle ne saurait etre identique chez tous les animaux puisque tous n'accomplissent pas de la meme fagon leurs trois types d'actes vitaux. Peut-etre est-ii done preferable de reperer, a l'interieur de chaque espece animale, comment se manifeste, clans l'accomplissement des trois types d'actes vitaux, des actes dus a la phronesis. Si cette phronesis renvoie bien a un certain type de recherche du bien participant quelque peu a !'experience et a l'habitude, le tout pouvant etre renforce par !'education, ii n'en reste pas mains que c'est de fagon differente que les divers animaux seront dits phronimoi. Non seulement ils accomplissent differemment leurs actes vitaux, mais encore ils ne sont pas egalement intelligents clans ces trois types d 'actes. De la vient sans doute que ce que l'on peut indifferemment nommer phronesis ou synesis soit souvent designe a !'aide d'autres termes specifiant plutot les qualites propres a tel ou tel animal. Les oiseaux seront ainsi volontiers qualifies d'eumechanoi et d 'eubioioi, qualites qui semblent communes a tous (H.A ., IX, 16, 616bl3-14)24, parce que leur « ingeniosite» clans la construction du nid ou dans la recherche de la nourriture leur permet de « bien-vivre ». D'un autre cote, l' intelligence des insectes sera qualifiee d'ergatike parce qu'elle est principalement ordonnee autour du minu­ tieux <( travail)> qu 'ils accomplissent, souvent ((en commun » comme chez les abeilles et les fourmis, ce qui les rend de surcroit politika (H.A . , I, I , 488a7-10) sans que ce soit pour autant des animaux aux sentiments et comportements familiaux developpes, ce .qui est le cas des quadrupedes (qu'ils soient gregaires ou solitaires) et, clans une moindre mesure, de beaucoup d'oiseaux. Ajoutons encore que de nombreux poissons sont qualifies de fechnika parce qu 'ils deploient de veritables ruses pour chasser, pour se defendre, ou bien encore pour se liberer de l'hamegon. Je voudrais maintenant, a !'aide de quelques exemples, developper certains de ces points afin de caracteriser un peu plus le type d'intelligence pratique a l'ceuvre chez divers animaux quand ils

(23) II faudrait d'ailleurs y ajouter les considerations relatives aux lieux (chOrai, 1 1 , 543b23-31 ; kala lous lopous, V I I I , 28-29). Ces diverses considerations sont destinees a montrer que non seulement Jes animaux, mais plus encore leurs comportements different selon Jes climats et Jes sols. (24) II semble cependant y avoir quelques exceptions : certains oiseaux sont dits ami!chanOteroi (614b34) sans que Aristote precise plus. Peut-etre s'agit-il du chardonneret (akanlhis) kakobios (616b31), ou bien encore de « i'aigle aux ailes tachetees de noir» (perknoplero.i;, 6 I 8b31-619a3). H.A., V,

, I· :11 :.1I .· f!:·. : ,, .

•

424

J.

L. LABARRIERE

accomplissent leurs actes vitaux. A priori, l'acte vital dans lequel se manifeste le moins cette intelligence est celui de la reproduction (j'entends ce terme au sens strict et designe par la l'accouplement et la copulation). Aristote, sauf erreur de ma part, ne mentionne jamais un acte de phronesis a ce propos. C'est tout a fait normal si l'on se souvient qu'en G.A ., I, 23, 73l a24-b8, ii soutient que l 'acte meme de se reproduire est un acte de l'ame vegetative (lo phylikon) et est done une fonction commune a tous les vivants. II n'hesite d'ailleurs pas a ecrire que !'animal, lorsqu'il s'ac�ouple, « devient comme une plante (ginelai h6speranei phylon) ». La seule difference d'avec les plantes est qu'etant donne que les animaux sont doues d'aislhesis, ii s'y ajoute du plaisir (hedone, H.A. , V I I I , 1 , 588b24-30). Or ce plaisir est tel qu'il conduirait plutot a l'hybris qu'a la phronesis, ainsi qu'en temoigne la fureur lubrique qui saisit les animaux au moment du rut (H.A., VI, 18). Cependant, si l'on rapporte cet acte a sa consequence\ la mise au monde de lekna, alors, avant meme qu 'il ne soit a proprement parler question d'elevage, les choses prennent un tour quelque peu different. L'exemple le plus probant en est sans doute la difference entre ces deux genres d'ovipares que sont le genre des poissons et celui des oiseaux. Sans qu'il soit ici necessaire de prendre en consideration les faits relatifs a la plus ou moins grande fecondite des oiseaux (G.A ., I I I, 1 , 749a34750bl), remarquons que la reproduction des poissons ovipares, a !'exception sans doute de la silure (glanis, H.A ., VI, 14, 569a4 ; IX, 37, 62l a20-b2), ajoute a !'imperfection de ce mode de reproduction, celle particuliere aux poissons (Ieurs ceufs sont inacheves, G.A., I I , 1, 732b2" 7 ; 733al 7-32 ; 733b7-10) mais aussi celle de la chance. En effet, si ces animaux sont tres prolifiques, c'est parce que nombre de leurs ceufs deja pondus un peu n'importe comment par les femelles sont de plus detruits OU devores par les males en meme temps qu'ils les fecondent en les arrosant de leur laite (H.A. , VI, 13 ; G.A ., I I I , 4-5). Et je ne parle meme pas ici du sort souvent tragique des jeunes alevins devores par de plus gros qu'eux, quoique souvent ils doivent leur survie a cette petitesse : ils sont dedaignes par les grands poissons (H.A ., V I I I , 19, 602b2-5). Heureusement des lors que la nature, faisant toujours de son 'mieux avec ce dont elle dispose25, compense en quantite ce qu'elle ne peut donner en qualite (G.A ., I I I , 4, 755a l l-35). Nul doute qu'un tel phenomene soit a rapporter au fait que Ies poissons ovipares n'elevent pas leurs petits mais les laissent grandir, ce qui heureusement, la nature faisant encore une fois de son mieux, se deroule tres rapidement (H.A., V, 10, 543a29-31 ; G.A., I I I , 4, 755al l - 1 4). En revanche, les oiseaux, ayant a elever jusqu'8 un certain point

(25) Le finalisme artistotelicien n'a en effet pas grand chose a voir avec une providence omnipotente, er. P.A., I I , 14, 658a23-24 ; IV, 10, 687a l 5 ...

1 I

DE LA PHRONESIS ANIMALE

425

leurs petits, ne pondent pas « a l'aveuglette » et manifestent a cette occasion de la « prudence )). En temoigne meme le contre-exemple du coucou (kokkux) qui, en raison de sa legendaire Iachete (deilia), agit prudemment (phronimon) en deposant ses ceufs dans le nid des autres (H.A . , VI, 7, 563b29-564a3 ; IX, 29, 6 18a25-30). On pourrait en dire de meme de la pigeonne (he perislera) qui, en cas de danger, peut s'empecher de pondre et retenir son ceuf (H.A ., V I , 2, 560b21-25). Mais le cas le plus probant est sans doute celui des perdrix (perdix) ou des cailles (orlux) auxquelles Aristote consacre un long chapitre (H.A ., IX, 8). Voila des animaux lubriques26, mechants et ruses (kakoethes kai panourgon, 613b23) et pourtant les femelles deploient une grande ingeniosite pour pondre et couver leurs ceufs : oiseaux au vol pesant et ne faisant en consequence pas de nid (613b6-12), elles ne pondent pas toujours au meme endroit afin que leurs ceufs ne soient decouverts ni par les chasseurs, ni par les males toujours prets a briser les ceufs afin de cacher a nouveau (613bl5-21, 25-27). Elles s'ingenient a trouver des parades contre ces ennemis, ce qu'Aristote souligne en qualifiant la femelle d'anlimechanomene. Ainsi par exemple, si pressees de pondre elles ne peuvent plus se retenir, elles pondent un peu n'importe oil, puis elles s'empressent de s'eloigner de leurs ceufs et vont parader devant les males OU les chasseurs afin qu'ils ne trouvent pas Jes ceufs. Elle agit de meme avec sa « nichee ». Nous avons done bien la quelque chose comme un raisonnement relevant de la ruse et de la « machination», ici fort poussee ainsi qu'en temoigne le terme anlimi!chanOmene, phenomene etranger a la ponte des ceufs chez les poissons quand bien meme ces derniers pondent dans des lieux propices (H.A ., V I , 13, 567bl 1- 18). II va de soi, ainsi que je l'ai deja mentionne a propos de la biche, que ce phenomene s'accentue des lors que nous avons affaire a des vivipares. Citons en ce sens les femelles des bisons (bosanos) qui se regroupent dans les montagnes afin de mettre bas, non sans avoir auparavant construit autour d'elles une sorte de rempart (peribo"lon) a !'aide de leurs excrements (H.A ., IX, 45, 630b l4-16)27• Les deux autres types d'actes vitaux, l'elevage plus ou moins long des petits et la recherche de la nourriture, sont bien entendu ceux dans

(26) C'est un point sur lequel Aristote revir,nt souvent : H.A., I, 1, 488a4 ; VI, 9, 564al l-12; G.A., I I , 7, 746bl ; I l l , I , 751a20 ... (27) Les bisons ont la reputation de 18.cher beaucoup d'excrements. Le font-ils simplement sous !'empire de la peur comme beaucoup d'animaux (P.A., II, 4, 650b30-33} ou bien, telle la seiche, le font-ils « intelligemmentf>? Outre que Jes bisons ont sans doute un sang de la meme nature que celui des taureaux et des sangliers, et sont done plutOt impetueux que craintifs, la deuxieme solution semble preferable, car Aristote precise qu'ils s'en servent comme d'une defense (P.A., I I I, 2, 663al3-I7) et que lorsqu'ils Jes projettent pour se defendre, ils ont alors la particularite de brUler, ce qui n'est pas lorsqu'ils se satisfont tout bonnement (H.A., IX, 45, 630b7-13).

1

i

426

I

'

i

ii ;i:

J.

L. LABARRIERE

lesquels Jes animaux manifestent le plus leur « intelligence pratique ». lei plus encore qu'ailleurs ii y a lieu de parler plutot de phroneseis. Par exemple, le mode de reproduction des insectes est considere par A,ristote comme le plus imparfai� qui soit (G.A . , I I , I), ce qui devrait conduii'e d'apres G.A . , I I I, 2, 753a7-l 7, a Jes considerer comme peu intelligents s'il est vrai que I' intelligence croit avec la plus ou moins longue attention portee aux petits. Or, nous savons bien qu'il n'en est rien et que l'abeille ainsi que d'autres animaux de ce genre est consideree carp.me plus intelligente que nombre d'autres animaux28• He bien ! Que cette intelligence soit systematiquement qualifiee d' ergaiike, de « laborieuse )>, dans l'H.A., IX, 38-40, me semble j ustement prendre en consideration ce fait : quand bien meme l'abeille, genre extraordinaire et divin (G.A., I I I, 10, 760a4-6, 76l a5), montre la savante organisation de la nature dans sa reproduction, elle n'est pas pour autant qualifiee d'intelligente eu egard aux « soins)) qu'elle prodigue aux larves, mais eu egard a l'ceuvre qu'elle accomplit dans une ruche organisee par la division du travail (H.A., IX, 40, 625b l 8-27). Son intelligence « laborieuse » et « epargnante » (623b 1718, 627a l 9-20 ; l'abeille partage cette qualite avec la fourmi, 622b26) n'est done pas de meme nature que celle des sanguins et des vivipares. Autrement dit, Jes animaux peuvent manifester de !'intelligence dans un domaine et pas dans un autre. II reste cependant vrai que ces traits d'intelligence ant, d'une fat;:on OU d'une autre, toujours affaire 3. leur sOteria, mais ils renvoient a l'assomption intelligente et non pas seulement mecanique de cette tache. C'est sans doute pourquoi Aristote etudie si longuement les oiseaux, blen qu'a suivre le jeu des comparatifs ils soient censes etre moins intelligents que Jes vivipares. S'ils sont qualifies d'eumechanoi et d'eubiotoi c'est parce qu'ils assument la tache de se conserver avec ingeniosite, ce qui rend leur vie sinon « bonne )} du moins aisee a vivre. Aussi bien dans la fagon de construire leur. nid, d'elever leurs petits en Jes accoutumant progressivement a leur future vie d'adultes, voire de se procurer leur nourriture (t3.che plus aisee pour Jes non carnivores, G.A. , I I I , 1, 749b24-25), on Jes voit deployer un art fonde sur one certaine experience et sur une certaine habit1:1de acquises par l'enseignement (H,A. , IX, 7, 612b21-3 1 , pour l'hirondelle, chelidon). Quel que soit le type d'aliments dont se nourrissent Jes oiseaux, Aristote soutient qu'ils ont un « raisonnement sUri> (ten tes dianoias akribeian, 612b20-21) parce qu'ils agissent fechnikos (IX, 13, 6 16a4) et sont par la eumechanoi, qualificatif portant aussi bien sur la fagon de construire le nid, de proteger et d'elever Jes petits, que sur celle de se procurer de la

(28) Bien qu'extraordinaire, le mode de reproduction des abeilles est imparfait. Comme celui de tous Jes autres insectes, ii n'entraine pas d'elevage a proprement parler. Aristote note cependant que Jes ouvrieres prennent soin des petits (G.A., I I I , 10, 759a36, b6-7), c'est-3.-dire qu'elles veillent au bon developpement des larves.

I

I

I

DE LA PHRONESIS ANIMALE

427

nourriture et de « sauver sa peau ». En voici un nouvel exemple tout a fait remarquable : « On dit aussi que les pigeons reconnaissent (ginOskein) chacune des varietes de faucons (hierax}, au point qu'au moment oil les faucons fondent sur eux s'il s'agit de ceux qui attrapent leur proie au vol, ils restent 13 oU ils sonl poses, et qu'au contraire, si l'attaquant est un de ceux qui foncent sur le sol, ils ne l'attendent pas et s'envolent . » (H.A . , IX, 36, 620a29-33).

Outre qu'une telle pratique demande un reel « sang-froid », elle in;iplique surtout quelque chose comme un raisonnement en forme de si . . . alors, raisonnement impliquant en amont que les pigeons soient capables de « classer » Jes faucons, et done qu'ils soient a meme de reconnaitre le genre faucon parmi les oiseaux a serres recourbees (ta ganpsonucha) puis d'y discerner ses diverses especes. II me semble done que la question de la phronesis animale gagne beaucoup si l'on cherche a en specifier ses aspects relativement au « bien » de chaque animal. II n'en reste pas moins que Jes choses ne deviennent pas absolument claires pour autant. Ainsi, s'il est permis de se demander si ce ne serait pas en raison de l'art que les oiseaux a serres recourbees ont a develop per pour se procurer leur nourriture, qu 'ils sont bien moins doues de l' intelligence' gregaire, voire familiale29, la meme consequence ne semble plus valable pour des vivipares carnivores tels lions ou dauphins (H.A . , IX, 44 et 48). Remarquons a ce sujet que !'intelligence des vivipares, Jes plus parfaits des animaux du point de vue de leur reproduction, n'est' pas traitee de fagon toujours satisfaisante par Aristote. D'une part I'epaisseur du sang de certains vivipares les rend moins intelligents que d' autres vivipares au sang leger ou que d' autres animaux dotes d'un analogue de sang leger, mais d'autre part la plus grande intelligence de certains les rend plus « familiaux », seraient-ils sauvages et solitaires. Or le rapprochement entre H.A . , V I I I , I , 589a l -2, ou Jes plus intelligents sont dits « plus politiques » parce qu'ils vivent plus longtemps avec leurs petits, et G.A., I I I, 2, 753a7-17 qui emploie le meme argument en soutenant que ces animaux ce sont les hommes et certains quadrupedes, sans toutefois Jes qualifier de « politiques», invite a se demander s'il n'y aurait pas un sens politico-familial de la phronesis ou synesis, sens caracterisant plutot Jes quadrupedes et certains vivipares, mais sens apparemment mal maitrise par Aristote lui-meme. Toujours est-ii que c'est ce sens la qui semble surtout interesser Aristote lorsqu'il parle des quadrupedes. J'en veux pour preuve le respect qu'a !'elephant pour sa femelle une fois qu'il l'a rendue grosse (H.A . , V, 14, 546b l 0 ; IX, 46, 630b21-22), ou bien le refus des jeunes chameaux de

(29) Aucun oiseau a serres recourbees n'est gregaire (H.A., I , I , 488a5) et de plus Jes aigles ne passent generalement pas pour des parents tres attentionnes (JI.A., VI, 6 ; IX, 32, 6!9a27-3! ; 34, 619b27-34). ,

I \�

\ ' '

428

B1otoom, r.oo1oul! er MtiAPHYSIOUE

J. L. LABARRIERE

monter leurs meres (H.A . , IX, 47, 630b31-63l a.l), ou bien encore !'extreme sollicitude des dauphins a l'egard de leur progeniture (H.A . , I X , 48). Ma conclusion sera done aussi simple que breve. Si nous ne pouvoris echapper a cette question : qu'est-ce que la phronesis animale ? et la traiter d'abord en nous demandant quel type de conna1ssance elle implique, ii semble qu'il ne faille pas s'abuser par la legitime question de l'interprete. En effet, en suivant la demarche aristotelicienl!e de plus pres, nous sommes conduits, semble-t-il, a substituer a� probleme de _ la phronesis celui des phroneseis. Cette approche, qm ne peut faire I'economie de la premiere, montre que par phronesis ou synesis Aristote n 'entend pas necessairement des qualites indifferenciees, et qu'au contraire ii s'attache plutot a specifier differents types d 'intelligences animales a l'ceuvre dans les fagons differentes d'accomplir les actes vitaux relatifs au bien propre de chaque animal.

chez

ARisrom

8eminaire CNRS-N.S.F.,

1987

Editions du CNRS, Paris, 1990.

SUBSTANCES AND SPACE-TIME: WHAT ARISTOTLE WOULD HAVE SAID TO EINSTEIN T1M MAUDLIN

Like Empedocles' �ouyov� &v8p67tpCJlp�, this essay will doubtless appear to be an ungainly, if not monstrous, concatenation of diverse topics. To conjoin exegesis of Aristotle's Metaphysics with an examina­ tion of the general theory of relativity must seem an act both ahistorical and philosophically perverse. And indeed, the marriage of these subjects is not an entirely happy one. The detail required for a satisfactory presentation of the textual analysis surpasses the needs of the later discussion of space-time, and so compromises the unity of the whole. Still, the advantages a fforded by this peculiar conjunction outweigh the drawbacks, for it illustrates the relevance of longstanding philosophical analyses to modern problems, problems which often present themselves in a gaudy technical garb. Although Aristotle had no notion of anything like modern physics, some of the puzzles he grappled with are exactly those which we face in trying to interpret our scientific theories. I hope to show that some of the recent debates concerning the status of space-time in general relativity afford a case in point. Aristotle's Metaphysics is a tract concerned with being, and the central books focus on those entities which most unqualifiedly partake in being: substances. Book Z, in particular, is devoted to the articulation of criteria by which to distinguish substances from non-substances; to the identification of substances; and to examinations both of the relationship between substances and their properties and of that between form and matter. This last topic is particularly pressing for Aristotle, for having supplemented the doctrine of the Categories with a hylomorphic ani!lysis of substance he must now determine whether and

430

T. MAUDLIN

how form and matter can combine to produce a truly unified substance. He finds that assimilating the relationship of matter and form to that of substance and accident has distastrous consequences, so he must struggle to find a new model for their inter-relationship. The . first section of this essay examines one of the central passages m the treatment of matter and form, the famous "stripping" argument of Z. 3. I shall contend that this passage contains a premise, commonly overlooked in discussions, which identifies the argument as non­ Aristotelean and whose denial provides Aristotle with a means of avoiding some absurd consequences. In order to explain why this premise is unacceptable, he must introduce the notions o� €:\lepyela (actuality) and Mvaµt<; (potentiality), using them to explicate the connection between form and matter. Form and matter are indeed, on Aristotle's account, much more intimately related than are substance and accident. This moral shall be employed in the analysis of the ontological status of space-time in the later sedions of the paper . . . The notion of substance has played a leading role m the ongomg debate about the nature of space and time. Although neither Isaac Newton nor Samuel Clarke claimed that space is a substance, the positions taken in the famous debate between Leibniz and Clarke have and relationism as respectively, known, be to come substantivalism. Various philosophical views asserting and denying that space and time are real entities distinct from physical objects have been refined, and the pendulum swings between them have often been guided by technical results of 'Current physical theory. One such argument has recently been presented by John Earman and John . . must be w1lhng to accept a Norton, who maintain that a substantivahst radical indeterminacy in physical law as the price for that ontology.' The second part of this paper will examine their argument and the strictures on substances that underlie it. I shall argue that the moral found in Aristotle has application here, since an Aristotelean account of the relationship between space-time points and their properties may provide the substantivalist an escape from Earman and Norton's dilemma. I shall also argue that the structure of the mathematical entities used to represent space-time naturally suggests a misleading analysis of the ontology of space-time itself, which analysis underpins the prima facie indeterminism in the substantival interpreta­ tion of the General Theory of Relativity. But let us begin with our passage from the Philosopher.

(1) The arguments are presented in Earman ms., Norton 1987, and Earman and Norton ms.

T

SUBSTANCES AND SPACE-TIME

43!

PART I: THE Z. 3 PuzzLE2 In Metaphysics Z. 3 Aristotle begins his investigation into substance. Four candidates for "the substance of each thing" are put forward: the essence, the universal, the genus, and the subject, the last of which is the topic of Z. 3 (I 028b34-36). The subject (�o OJtoxdµevov, literally "what underlies") is identified as that of which everything else is predicated, and which is itself predicated of nothing else. This characterization immediately rules out all entities from the categories other than substance from being the ultimate (7tpw�ov) subject since qualities, quantities, places, etc. are always qualities, quantities and places of something. Yet an individual such as Socrates, of whom qualities such as musicality may be predicated, and who would be a paradigm substance in the analysis Aristotle gives in the Categories, can also be analyzed into form and matter, and so it is not yet clear whether the ultimate subject is the form, the matter, or the compound of the two (1029a2-9). We must therefore examine which of these has the best claim to be that of which everything else is predicated. At 1 029a7-30 Aristotle presents one way of understanding what a subject is, a way which entails that the identification of substance with the ultimate subject must be mistaken. The argument runs:3 So it has now been stated in outline what substance is, viz., it is that which is not said Of a subject but everything else is said of it. Yet we must not only define it thus, for this is not sufficient: it is unclear, and moreover matter becomes substance. For if it is not substance, whatever else might be escapes us; for since everything else has been stripped off nothing seems to remain. For all the other things are, on the one hand, affections and products and powers of bodies, and again the length and breadth and depth are sorts of quantities and not subst�nces (for quantity is not a substance); but, rather, that to which these ultimately belong, lhal is substance. But since length and breadth and depth have been stripped off, we don't see anything left unless that which is bounded [or defined] by these is something, so that matter alone must seem to be substance to those who investigate it in this way. I mean by "matter" that which in itself is said to be neither something (•�)4 nor so-much nor any of the other things by (2) I am indebted for very useful comments on the section to Rob Bolton and especially to Mary Louise Gill, whose influence permeates the whole. (3) All translations are my own, from the text of Jaeger. I have aspired to literalness rather than elegance and have tried to maitain consistent translations of the central technical terms. (4) The translation of the term "·n" throughout this passage cannot but be tendentious. Aristotle says that according to this argument '"n" cannot be predicated of the ultimate subj �ct, but then goes on to use that term in referring to it. This might

11

I ' I

" '· .

432

! :

I

:I !

' 11. '.

L.1 ' ·' ,, ii

T. MAUDLIN which being is defined. For there i� sof!let�ing of which each ? f these is . predicated something (·n) whose being is d1st1nct from the being of the things pred icated (W "t"0 el'JIXL �Tepo'J xixl -rfu'J xix"t"eyopLfu'J Ex
On the face of it, this passage presents a reduclio ad absurdem of the definition of substance as that of which everythmg is predicated while it is predicated of nothing else. For the definition seems to pick out as substance some sort of matter which of itself 1s entirely characterless, indefinite and amorphous; but substances are pre-emmently md1v1dual things (cf. 1028al 1-20). But, as with every aporia which Aristotle presents, we must try to determine how much of the argument is m propria persona, whether Aristotle is committed to_ each of the premises, and whether there is any escape from the conclus10n. In th1_s mstance such an inquiry is especially urgent, for if the argument is m propria persona, not only will it show that Aristotle believed the ultimate . subject of predication not to be substance, it will also md1cate that he _ _ regarded that ultimate subject as entirely characterless. That is, this passage might seem to prove Aristotle to be committed to the doctrme of prime matter.5 . . . I do not mean to canvass the arguments for and agamst prime matter here, but a few remarks on the significance of this passage are m _ order. The argumentation of this passage arrives at prime matter m a way quite distinct from the other texts which are sometimes taken to establish that doctrine (passages such as de Gen. el Cor. 329a24-35 . and 332a34-b I). Those arguments, which deal with the cond1t10ns necessa-

suggest that the term is being used ambiguously; .R� bert Bolto� has su�ge�;ed to. me that whereas here it means "a particular thing", later 1t JUSt means something (albeit �ot a� member of any of the categories). This interpretation can remove some se\f-contrad.ict�r1ness from the passage, but to so translate it would hide the (at least) surface parad.ox1cahty of the view. I choose to translate in uniformly as "something" because I believe t� at paradoxicality to be intended. An exactly parallel, and evidently intent.ion � !, rhetorica � flourish is used by Plato in a text which I believe to be one of the sources of this argument. Theaelelus 202a4. More on that passage anon. {5} Whether the passage is in Aristotle's own voice is already a matter of son:i e debate. W. Charleton, for example, takes the main line of argument to be that of certain opponents of Aristotle who are identified in Physics .11 {Aristotle 1970, p . . 138). II. M. Robinson, in contrast, states that he finds no plausible grounds for supposing a change in voice (Robinson 1 974, p. 184} . . Much o� this �art of the essay will be devoted to defending the vie\\' that the argument 1s not Aristotle s.

SUBSTANCES AND SPACE-TIME

433

ry for the elements to change into one another, depend cruciall y on certain empirical observations. Were the elements not to be mutab le, there would be no questio n about what, if anything, remains through out that transformation. Our passage, however, is metaphysical or logical rather than empirical or physical. It depends not upon any observe d physical behavior of substances but only on the nature of predication. Thus, although the entire issue of prime matter would not be resolved should this argument not be in propria persona, still one distinct approach to that doctrine would be blocked , and the relevan t evidence would be confined to the analysis of the transmutation of elements. I shall maintai n that the argument of Z. 3 is not Aristotle's own, that he is not committed by it to the doctrine of prime matter, and that it employs what is for Aristotle a mistaken account of the relationship between form and matter in a compou nd substance. · I shall begin by presenting some circumstantial clues that suggest this reading . But the major burden of this part shall be an exposition and analysis of the premises employed in the argument. Among these we shall find one that runs counter to the ensuing discussions in the Metaphysics, which allows us to see just why Aristotle need not accept the conclusion of Z. 3. Finally, I shall suggest a possible source for the Z . 3 puzzle. But let us begin with some superficial evidence. There are clear indications in this passage that Aristotle was following out a line of thought already extant in the philosophical literature. He notes that various things follow "for those who investigate it in this way" (1029a1 9, cf. also h µev oilv wuTwv 6ewpoi)cr, at a26). This may only refer to who accept that the ultimate subject is substance, but it may also indicate that other peculiar premisses are used in the deduction. The conclusion of the passage seems stronger than the announced intention and also conflicts with Aristotle's later p"onouncements. For although he begins by suggesting that the characterization of substance as subject is obscure and in need of clarification, he seems to end by rejecting it altogether since it implies that a totally indeterminate matter is substance. Yet the synopsis and review of Z which begins chapter H still recognizes the subject as substance (1042a 13) and identifies this underlying substance as matter (1042b9). So either Aristotle has simply forgotten the result of Z. 3 or he feels that those objections to the subject-criterion of substance have been overcome. Let us take the latter possibility" as a working hypothesis and seek a premise of the Z. 3 argument which may have come to be rejected . If the foremost point of the Z . 3 passage is to identify the ultimate subject of predication, one line alone seems sufficient to solve that problem: "For all the other things are predicated of substance, and this

I

\I T. MAUDLIN

SUBSTANCES AND SPACE-TIME

itself is predicated of matter" (1 029a23). Aristotle nowhere denies this assertion, and it corresponds with the identification of substance-qua­ subject with matter in H. I and H . 2. But it is not simply the result that matter becomes substance which generates the difficulties of Z. 3, but that the matter should be so entirely characterless, possessing none of the usual marks of substance. This reduction of matter to the totally indefinite does not follow simply from the claim that all things are predicated of substance and it of matter. Abstraction of all the predicates from the subject is involved, this being the strippi n g process so central to the text. What is involved in this abstraction ? What conditions must hold for the separation of predicate and subject to be possible? The stripping process of Z. 3 is a matter of logical abstraction, of disregarding properties. Just as we construct a proof about triangle qua triangle not by considering a triangle which is neither scalene nor isosceles but by allowing the proof only to employ those features mentioned in the definition of a triangle, so too can we more generally disregard the inessential features of an object. We do not abstract color, say, from Socrates by imagining him to have no color, or by considering processes for removing all color from him, but by recognizing that the characterization of Socrates qua human being need not advert to his color.6 If the stripping process is such a disregarding of properties, then the success of the procedure critically depends upon one condition: the predicate disregarded cannot be part of the defining conditions of the subject of which it is predicated. The essence of the predicate and that of the subject must be logically distinct, else the result of disregarding the predicate will not be to leave one with the subject as such. The logical independence of the predicate and subject, then, is an essential premise of the Z. 3 analysis. Further, it is not employed by Aristotle as a tacit background assumption, but is quite explicitly stated: "For there is something (�•) of which each of these is predicated, something whose being is distinct from the being of the things, predicated" (1029a21-23). Indeed, this premise in conjunction with the assertion that "all the other things are predicated of substance, and this itself is predicated of matter" immediately yields the conclusion that matter

cannot be substance. For whatever characteristics make something a substance are predicated of matter, which in itself must have a distinct, and hence non-substantial, essence. So our quarry is nearly cornered: we now have a simple two-premise argument whose conclusion conflicts with Aristotle's retention of matter or the subject as a sort of substance in book H . If he is to be consistent he must either reject that su_bstance is predicated of matter or that subject and predicate always have logically distinct beings. The distinctness of the being of the predicate from the being of the subject is prima facie a very questionable premise. After all, one would suppose that anything of which substance is predicated thereby is a substance, rather than the opposite. And, to take a mundane example, although we all agree that there is something of which the hardness, rectangularity, color, etc., of the table are predicated, we would balk at the suggestion that the subject is something entirely distinct from these properties. After all, it is the table which is colored, hard, rectangular, etc., and we have no reason to believe that the table could be defined (or could exist) independently of these qualities. The distinctness thesis immediately implies that nothing can be affirmed of the ultimate subject of predication as such (x0<6' "'��o), for to do so would be to predicate something of it. Indeed, despite his sanguine locutions early in the passage, Aristotle is forced to forgo calling the ultimate matter even something (n) (1 029a20). As the reader will doubtless have anticipated, it is the premise which asserts the universal distinctness of the being of the subject from that of the predicate which I believe not to be accepted by Aristotle. I f h e has imported this premise from elsewhere, w e should expect him to develop an account of predication which denies it. And should this premise fall, with it would fall both the general argument for prime matter in Z. 3 and the objections to the characterization of substance as the ultimate subject of predication. My evidence for this reading of Z. 3 will be of three types. First is a text which directly indicates that Aristotle rejected the Z. 3 picture of predication. Next we shall examine Metaphysics Z and H to see whether Aristotle's considered views allow him to avoid the aporia. Finally, a possible non-Aristotelean source for the Z. 3 argument will be suggested . The stripping procedure o f Z . 3 does not immediately strike one as non-Aristotlelean because the process of abstraction is one Aristotle often uses. One commonly can disregard various predicates of a subject, and must do so to properly understand qua what the subject has certain of its properties. And the stripping process clearly is legitimate when the being of the subject and the predicate are distinct. But this holds just when the predicate is an accident of the subject, rather than

434

(6) Russell Dancy, in [Dancy 1978], assimilates the abstracting process to one or thus obscuring the vital difference between this logical argument and the empirical arguments concerning substantial change. Dancy suggests, for example, that we "strip off" a statue's color using turpentine {p. 398). Such appeal to physical procedures is clearly too weak for Aristotle's purpose: how could one strip off color generally, or be assured or an actual process which would eliminate all of the particular affections, products, and powers of bodies? imagining the properly physically removed from the substance,

435

437

T. MAUDLIN

SUBSTANCES AND SPACE-TIME

part of its essence. The Z. 3 thesis, then, is that every predicate is an accident of its ultimate subject, prime matter, and hence can be logically stripped from it. But Aristotle denies this very thesis in Metaphysics r. 4:

Metaphysics 1037a5 ff., 104l b6). The immediate matter of a human is a complete body composed of non-uniform parts such as the face, hand, heart, etc. The matter of these in turn are the organic uniform parts such as blood and bone.8 What is the logical relationship between the being of the form and the being of the matter in this case.? Aristotle is quite clear and emphatic on this point. The non­ uniform parts, and even their uniform constituents, only are what they are when ensouled. A dead hand is a hand in name only, like the stone hand of a statue. A finger cannot survive separation from the body and remain a finger. Even flesh and bone, although their activities are less immediately evident, are truly flesh and bone only when part of a functioning animal (cl. Metaphysics !035b22-27, 1036b27-32, 1 040b5- 16, de Gen. An. 734b25 ff.). So in the case of ensouled beings, the matter of which the soul is predicated only is what it is when ensouled. This dependence is reflected in the definitions of the parts: the parts are logically (and hence ontologically) posterior to the whole since their definitions must make reference to the whole ensouled organism:

436

And those who say this generally destroy the substance and the essence. For it is necessary for them to say that everything is accidental... To indicate the essence [of something] is to say that its being (-rO eivoc� ocUT<;'.>) is nothing else... So that it is necessary for them to say that there will not be such an account [i.e. of the essence] of anything, but all attributes are accidental. For substance and accident are distinguished by this: the white is accidental to the man because although he is white, white is not in his essence.7 But if everything is predicated accidentally, there will not be any ultimate thing of which they are predicated, if the accident always indicates a predicate of some subject. For it [i.e. the identification of new subjects] must go on ad infinitum, but this is impossible (1007a20bl).

The argument refuted in this passage bears a strong resemblance to that of Z . 3. Someone has taken a position which implies that the predicate is always accidentally predicated of the subject, that it is never part of the essence of the subject, so that the beings of the predicate and of the subject are always distinct. But then the being of the subject itself must be predicated of yet another distinct subject, and so on indefinitely. The Z. 3 passage bites off the regress by positing an ultimate subject which has no essence, nothing true of it as such. But Aristotle would surely conclude that the Z. 3 argument, as much as this, destroys the substance and the essence since according to it nothing is ever essentially predicated of a subject. If Aristotle here rejects the notion that all predication is accidental, it is difficult to imagine that he would accept in Z. 3 that the ultimate subject of predication is prime matter, all of whose positive attributes at a time are accidents. If, then, the Z. 3 argument is not in Aristotle's own voice, and if the unacceptable premise is not the identification of substance with the ultimate subject but rather the thesis that the being of the predicate and that of the subject are always logically distinct, then we should expect Aristotle to develop an account of the relationship between substantial form and the matter .of which it is predicated which contradicts that thesis. This is just what he does. Let us begin by considering the most clear and, prima facie, undeniable example of a substance: a living animal. An animal, such as a human being, has a form: its soul. That soul is predicated of a certain particular kind of matter: an organic body (cl. de Anima 4 1 2a28,

(7) The phrase here, oUx 0Tte:p Ae:ux.6v, does not explicitly mention the essence, but is besl rendered as saying that white is not in the essence of man. Cf. Jonathan Barnes commentary in his translation of the Posterior A nalytics (Aristotle 1975) p. 168.

And the account of the right angle does not resolve into the account of the acute, but that of the acute into that of the right. For someone defining the acute must make use of the right since "acute" is "less than a right angle". And the circle is similarly related to the semi-circle, for the semi­ circle is defin€'.d by the circle, and also the finger by the whole; for a finger is such-and-such a part for a man. So whatever are parts in the sense of matter, into which a thing is resolved as into matter, are posterior... (1035b6-12).

Similarly, the entire organic body of which the soul is predicated must be logically posterior to the soul since without the soul ii wouldn't be the sort" of body it is. Definition of the body must make mention of the soul and its functions. It is perhaps apposite to remark here that there is every reason to believe that Aristotle would have considered the ultimate form, the species, of an animal to make the matter peculiar to that species. There is sometimes a tendency to think of different sorts of animals as being made of flesh and blood in the way cups and saucers can be made of clay, as if only the arrangement of the parts differentiates the species. But is anything so obvious as that horses are made of horse-flesh and humans of human flesh? There is clearly an intrinsic difference between beef and pork, and I see no reason to believe that Aristotle would have thought that a pig could be made of the former and a cow of the latter. Down lo the species level the form makes the matter what it is.

(8) The non-uniform parts are those whose own parts are not of the sa1ne sort. Thus, a part of a face such as a nose is not itself a face. In contrast, a part of flesh is itself flesh and a part of bone, bone.

439

T. MAUDLIN

SUBSTANCES AND SPACE-TIME

It might also be remarked that this definitional dependence of the matter of an animal on its form is a stronger relationship than that of hypothetical necessity. It is not just that the performance of the . psychic functions requires a certain kind of matter, as the functwn of a saw requires it to be made of iron (Physics ! 99b34-200a l 4). The strongest possible case of hypothetical necessity would be that where the form or function of an object entailed precisely the sort of matter from which it must be made, from iron, say, or such-and-such a sort of iron. But even in this case, the being of the matter would riot depend on the form: it is no definitional part of being iron, for example, that it be made into a saw. In the case of an animal, it is not just that the form requires a particular sort of matter to be realized, but that the matter essentially requires a particular kind of form. This dependence of the subject on its predicate in the case of predicating a substantial form of matter permits Aristotle to escape the z. 3 aporia. Since the being of the matter is not logically distinct from the being of the form,' the abstraction process cannot be carried out. Forrn and matter, in the case of composites such as animals, manifest the sort of unify which is the hallmark of true substances. Much of Metaphysics Z, H , and 0 is devoted to the explication of how such a unity is possible. In the case of the unity of matter and form in a composite, the problem is originally motivated by the Z. 3 puzzle. The definitional and ontological dependence of matter on substan­ tial form presents Aristotle with a new problem to solve. For he still wants to maintain that the two, as subject and predicate, are not identical. In the case of the part-whole relation this is easy to maintain: although "man", and hence "soul", must be mentioned in the account of a finger, it is clear that the finger is not identical to the, man, the part to the whole. But in the case of the relation of the whole organic body to the soul the problem of definitionally distinguishing the two becomes acute. A man can survive without a finger) and not vice versa, demonstrating the priority of the former to the latter. An animal body cannot, strictly speaking, exist without an animal soul, but neither can the soul · exist without the body.10 Aristotle is therefore

faced with a dilemma. On the one hand, he wants to maintain some difference, if only in account, between soul and body. On the other, any such distinction threatens to turn the unfied soul-body composite into two things rather than one. Furthermore, a parallel difficulty faces Aristotle when he turns to the form itself. Accounts of forms are often given by providing genus and differentia. Yet what is the logical relationship between these? Again, they clearly are not identical, but if they are two distinct things the unity of definition is threatened since the form defined by genus and differentia will become a plurality. More generally, all of the items cited in an account of a substance cannot themselves be substances without risking such a plurality, but neither can such an account mention only non-substances lest substances be constituted merely out of qualities (1038b23). Aristotle poses this difficulty and sketches its solution in Z. 13:

438

{9)

In saying that the being of the matter and that of the form are not logically distinct. I do not mean to imply that they are identical, just that one must make reference to the form in defining the matter. It is this degree of logical interdependence that thwarts the stripping process. (10} The ability to exist separately does not always accompany the part-whole relationship. Some parts, such as the heart and brain, are "simultaneous" with the body since it cannot live without them (1035b25-26). I will not here treat of active voU�, the one type of soul that seems not to require any body, not being associated with the exercise of a bodily part.

I

I

I

For it is impossible for a substance to be made out of substances actually inhering in it, for things which are actually two are never actually one. But if something were potentially two it might be one. For example the double is made out of two potential halves, for the actuality [of the halves] separates it. So if a substance is one it will not be made out of substances inhering in it whi�h also exist in this way, as Democritus rightly says. For it is impossible to say that one thing is made out of two, or two out of one . . . (1039a3-10).

The key notion alluded to here is the distinction between actuality and polenlialily. This pair of correlative terms is central to Aristotle's resolution of the problem of the unity of substance, and he devotes all of book e to it. The distinction is applied to all of the outstanding difficulties surrounding the analysis of definition and substance: the relation of genus and differentia; the relation of the parts of a definition to the object being defined; and our problem of the relation of form to matter in a composite substance. In fact, Aristotle often treats the problems of genus/differentia and form/matter as identical. This is not surprising: at 1 024b8-9, when defining y&'o' (genus), Aristotle mentions that in one sense it is a subject, which is called matter (cf. 1038a6-7, 1043b3 1). Let us consider the solution to the problem of genus and differentia, of why, if man is by definition two-footed and animal, he is not two rather than one. Aristotle writes: It is clear that if [certain Platonists] seek to define and to speak as they are accustomed it will not be possible to remove and solve the diffieulty, but if there is1 as we say, on the one hand the matter and on the other hand the form, and the former is potentially, the latter actually, the subject of investigation will no longer be taken to be a difficulty {1045a20-25).

A proper investigation of the actuality/potentiality distinction would take us too far astray,11 but some brief observations can be (11)

A detailed treatment of these issues congenial to this account has been given by

440

T. MAUDLIN

made. Jn Aristotle's resolution to the problems of unity, form is actual, matter potential, final differentia is actual, genus potential. Further, the connection between these relata in a substance is much more intimate than accidental predication. The substantial form of a composite makes the matter what it is; it is the actuality, the realization of the matter. 12 Houses, books, boxes, etc., all of which have separately identifiable matters (stones, beams, pages, boards, etc.) are not really substances for this reason, although they do have forms which account for their most important properties.13 This fundamental definitional unity of form and matter as potential and actual solves both the Z . 3 dilemma and the difficulties about the unity of a substance and of a definition: So the final matter and the shape are one and the same thing, the one potentially, the other actually. So to seek the cause of a unity is like seeking the cause of its being one. For each thing is some one thing, and the potential and the actual are in a way one . .. (1045bl7-21). Having shown that there are independent textual grounds for suspecting that the Z. 3 argument is not in propria persona, and having demonstrated that the argument relies on an explicit premise, viz. the logical distinctness of predicate and subject, which Aristotle elsewhere rejects in his account of form and matter, all that is left is to suggest a possible source for the Z. 3 puzzle. As this question is quite tangential to our main purpose, and the proposal made shall be highly conjectural, I shall only touch on the topic. Perhaps the most obvious place to locate the source of the Z. 3 puzzle is in the Timaeus. In that treatise space is introduced as a receptacle for the forms, in which they are realized only transiently and imperfectly. In order to receive the forms, the receptacle itself must be inherently characterless, like the matter of Z. 3 (Tim. 50d-5lc), This passage surely forms part of the background of Z. 3. But . if the

Mary Louise Gill in [Gill ms.]. I have drawn from her insights iqto the nature of composite substance in the following sketch. (12) The differentia similarly transforms and realizes the genus in each particular case. Aristotle says that the animality of a hors� differs from the animality of a man, the species differentiating the genus itself non-accident.ally (cf. Metaphysics I. 8 1057b32 ff.). If so, one can no more abstract the genus from the differentia in a given case than the matter from the form. Thus the genus as such is potential in an even deeper sense than just as being a determinable which has no existence apart from determinates. (13) The examples are given at 1042b12 ff., the denial of their being substances at 1043a4. 1 take cxl>1"1J Yi £..,kpye:�0t at 1043a6 to mean that in a substance the form is the actuality of lhe mailer of which ii is predicated (as opposed to just being the actuality of the composite). The form of a house is the actuality of the house, it is what makes the house be a house. But being-part-of-a-house does not make a stone what it is, while being-part­ of-an-animal does make flesh what it is. Not surprisingly, parts of animals and heaps of elements also fail to be substances (cf. 1040b5 ff.).

, SUBSTANCES AND SPACE-TIME

441 distinction between logical and physical arguments for prime matter is sound , the Timaeus must fall in the physical camp . Space is said to be characterless in itself because of the extreme mutab ility of the images engendered m it, because of the transmutation even of the elements into one another. The passage which leads up to the introduction of space focuses not on questwns of predication or definition but on the complete . mstab1hty of spatio-temporal objects (48c-50d). But we can find a less apparent source of the Z. 3 argument if we turn to another Platonic . wo_rk which wrestles with many of the same probl ems which confront Aristotle m the Metaphysics: the Theaetetus. Although the early sections of the Theaetetus, which deal with the . thesis that knowledge is perception, are not particularly relevant to the concerns of the Metaphysics, it is clear that the final section, from 20ld to the end, must have been before Aristotle's mind when he wrote the central books of his treatise. In this last part, Plato considers the . suggestwn that knowledge is true belief with an account (Myo<; ) of the thmg known. This raises the question of what sort of entities have the appropriate kind of account, and of the nature of an account, just the problen: s of defimtwn which Aristotle must face. Thus , for exam ple, Plato discusses whet her an objec t of knowledge can have parts, or be the sum of its parts, at 204a-206a, and Aristotle examines these same questwn m Z. 10-1 1 . More precise parallels can be found throughout . the two works. 1 4 This last section of the Theaeletus begins with Thea etetus and Socrates considering a new accou nt of knowledge, one which each has heard somewhere : I seem to hav�. heard some people say that the funda mental elements, as it were, from which �oth we are comp sed and everything else, don't have an ? cc � .ount (A6yoi:;): F or each of these 1n and of itself can only be named and �t � s not possible to say anything more, either it is nor th� t it isn t. For that would be to append "substance" orthat "not but one must attach nothing to it if one is to express substance" to itf that thing itsel � lo�e,., So one must not attach "that thing" or "itself" or "alone" or this or any of the other �an)'. such terms. For these run all about being . attach�d to �v� ryth1� g, be1n distinct fron: t at to which they are attach ,, ed (e:-re:pcx ov-r� e:x.e:Lvwv oti:; 7tpocr-r9L6e:-rcxL). But if �it were possib le for it to be d �fined (Ae:ye:cr6cxL) and to have an account peculiar to it, it would be defined without all of the othe� terms. So it is i m ossible to express any of the . :pis fundamental �lements 1n an account, for 1t not possib le to do anything but to name it alone . . . (20le-202b). This account shares essential features with the z. 3 argument. It . is logical rather than physi cal. It argues that the ultimate elements of any account must themselves have no account. Not even "substance"

(14) Compare, for example, Thealelus 208d with Metaphysics 1040a27-b2·' 205a w1· th 1043b36-44a2; and 203d ff. with 104 1 b l l ff.

'· :

T . MAUDLIN

SUBSTANCES AND SPACE-TIME

can be predicated (in Plato's terminology "7tpocrq>Ep
separate Forms. Plato would insist that any true substance must participate in the Form of Unity, and hence that the elements of the dream passage can be neither substances nor particulars. 1 6 To summarize this section, then, we have the following results. In Z. 3 Aristotle presents an argument which reveals a difficulty in understanding how physical objects can be substances. One usual mark of substance is being the bearer of predicates, but if we continually abstract the subject from the predicate in analysing a material object we shall ultimately arrive at an entirely indeterminate matter which seems not to be substance at all. I have argued that Aristotle does not accept the Z. 3 abstraction process as universally feasible. It conflicts with remarks he makes elsewhere. More importantly, it requires a premise, the universal logical distinctness of subject and predicate, which Aristotle will reject in his discussion of matter. The potentiality/actua­ lity distinction provides Aristotle with an account of how matter and substantial form can be unified in definition, thereby blocking the Z. 3 regress. Finally, a plausible source for the Z. 3 argument can be found outside Aristotle's works in the Theaetelus. Understanding Aristotle is a worthy end in its own right, but I have not presented this account purely as explication de texte. The Z. 3 puzzle is central to all metaphysical analyses concerned with physical substances. If one wants to identify substances as bearers of proper­ ties, the relationship between the predicate and the subject must be carefully considered. We must chart a path between the Scylla of bare particulars and the Charybdis of Leibnizean complete concepts if we seek to maintain that some, but not all, of the predicates of an individual' make it the thing it is. In the following section we shall apply this moral to a problem of contemporary metaphysics: the ontological status of space-time.

(15) As we have seen, the real trouble comes when we reach the point at which substance is predicated of matter and we must try to separate the matter' from substantial form. One might also trace out the other branch of the division a t this point, the predicate, and run a similar argument. This is just what Aristotle does at Metaphysics B. 5 100lb27 ff. By an identical process to that of Z. 3 Aristotle strips ofr all the affections, motions, relations, arrangements and ratios {)1.6yo�) of bodies as not being substance. But when considering whether surface is more of a substance than the body it defines (or bounds), Z. 3 opts for the subject, the indefinite extension, while B. 5 selects the bounding surface. This line of attack leads not to prime matter as the ultimate substance but to points and lines. But surfaces, lines, and points are divisions of bodies (1002al8), present in them only polenlially (1002a20), and so cannot themselves be substances. From Aristotle's point of view, the Platonic/Pythagorean analysis of physical entities into a combination of the (ontologically prior) limit and the unlimited, or the one and the indefinite dyad, yeilds the B. 5-Z. 3 dilemmas. For the unlimited is too indefinite to be substance, while unit measures are posterior in account to the substances of which they are measures (cf. Metaphysics N 1087b33 ff.).

PART 2: THE ESSENCE OF SPACE-TIME

442

443

Although the Scholastic metaphysics of substratum and substantial form, accident and essence, was resolutely rejected during the scientific revolution, fundamental questions of ontology, and hence of metaphy­ sics, cannot but be revived by changes in the foundations of our theoretical understanding of the world. Newtonian dynamics challen­ ged the mechanical world-view, special relativity rejected not only absolute space but the categorical distinction between spatial and (I 6) The differences between the Z. 3 passage and that of the Theaetelus were impressed upon me by Sarah Brodie, which is not to say that she will be satisfied by my attempt to establish their funda1nental philosophical unity.

T. MAUDLIN

SUBSTANCES AND SPACE-TIME

temporal properties, and general relativity replaced the static space­ . time continuum with one dynamic and non-(pseudo)-Euchdean. But mathematical formuli do not immediately yield up ontologies, and the most sweeping changes in theory are also the most subject to varied interpretation. We are still trying to come to grips with a theory introduced over 70 years ago: Einstein's general theory of relativity. Despite the striking mathematical clarity and beauty of the theory, it can be rather sibylline in its pronouncements abouL the ultimate furniture of the world. The special theory of relativity welded space and time into a unified space-time continuum which admits of no unique partition into spaces­ at-a-time. Still, as a 4 dimensional object, Minkowski space-time has a fixed universal geometrical structure. The primary insight of the general theory was to allow the geometrical structure itself, the space­ . time metric, to become a dynamic entity influenced by the d1stnbut10n of matter and energy in the universe. Einstein often motivated this change, in part, by an appeal to a sort of generalized version of Newton 's _ third law: since the trajectories of massive particles are determmed m part by the inertial structure of space-time, the inertial structure sho uld . itself be affected by the particles.17 This seems to elevate space-time and the metric into physical objects co-ordinate with material particles. Having sacrificed the magisterial aloofness from mundane events which the space-time of special relativity enjoyed, the space-time of general relativity becomes a dynamic physical entity. . space­ From this point of view, general relativity seems to establish time as a sort of physical substance, capable of entering into reciprocal causal interaction with the commoner class of objects such as tables and chairs. That Einstein so regarded it is suggested by his reference, in several articles and addresses, to the " aether" of general relativity." One is tempted to regard the space-time of general relativity as an immaterial rubber block, contorted and twisted by the matter in it. Still, nothing amounting to an argument has been

produced, and the mathematical entities introduced in a physical theory are always vulnerable to being analyzed away, shown to be fictions or abstractions which don't directly represent the physical reality. And recently John Norton and John Earman purport to have found an argument which demonstrates that if determinism is to have a chance, space-time cannot be a substance.19 It is this argument which shall be our focus in this section. Before presenting it, we must briefly review the mathematical structure of the general theory.2• Our project is to construct a mathematical object which represents the universe. We shall do this by beginning with mathematical simples and adding on successive layers of structure. One must keep in mind that the (abstract) ontological structure of the mathematical representa­ tion may suggest, but does not entail, an analogous metaphysical analysis of the physical structure that it represents. Indeed, one of my main contentions is that a failure to distinguish between the ontology of the mathematical representation and that of the thing represented has led to Norton and Earman's conclusion. We begin with a simple collection of points. The points are pure abstract individuals, bare mathematical particulars which, initially, bear no properties and enter into no relations. Still, as mathematical objects they are distinct individuals. If you like, you can call these objects space-lime poinls,21 but one must bear in mind that they do not as yet have any spatio-temporal properties and that nothing yet suggests that the set is particularly suited to represent space­ time. That is, as mathematical objects space-time points do not differ from any other kind of points - they only become connected with

444

(17) Cf ., for exa1nple, .Einstein (1956) pp. 55-56. That Einstei � took this ?e�eralized law of action and reaction seriously as a physical principle, and did not use it Just as a rhetorical device 1 is evidenced in his research into unified field theories. In 1929 he rejected a set of field equations proposed by Elie Cartan because he believed that they allowed a causal influence of the metrical structure of space on its par.allel structure without positing a causal connection in the other direction (Cartan and Einstein 1979, pp. 73-74, with responses on pages 81 and 89). . has (18) Three examples are cit.ed by Norton in Nort.on 1987, p. 183. Art.hur Fine stated t.hat. Einst.ein had earlier drawn exactly t.he opposit.e inference, t.hat. the general t.heory st.ripped space-t.ime of all physical realit.y (Fine 1984, p. 91). I argue in t.he appendix t.h at t.his impression is only creat.ed by taking cert.ain passag:s ou.t. o� cont.ext, where t.hey refer not. t.o t.he physical realit.y of space-t.ime but. t.o t.he physical significance of coordinalizations of space-t.ime.

445

(19) The papers in which t.his line is developed are Nort.on 1987, Earman ms., and Earman and Norton ms. Norton, in [Nort.on 1987] claims t.o find t.his argument. in Einstein's papers of 1913 and 1914. See t.he appendix of t.his paper for some observat.ions on Nort.on's claims. (20) Throughout. t.his sect.ion I have st.riven t.o avoid t.echnical detail, all.hough some familiarit.y wit.h t.he t.heory is presupposed in cert.ain of t.he foot.not.es and in t.he appendix. I have no quarrel wit.h t.he mat.hemat.ical det.ail of Earman and Nort.on's argument., only wit.h t.he ont.ological interpret.at.ion of it.. I hope my somewhat. int.uit.ive sket.ch capt.ures t.he fundamental structure of t.he argument simply and clearly. (21) There is an extremely unfort.unate ambiguit.y in t.he term space-lime points: sometimes it refers to element.s of the mat.hemat.ical object which represents space-time, and somet.imes to physical event.s. This ambiguit.y should be kept in min� when reading this paper and when comparing the Iocut.ions here t.o at.her works. I hope that. my int.ent.ion will always be clear in context. This confusion of mat.hemat.ical represent.at.ion with thing represented infects much mathematical physics. For example, it. is perfect.ly natural to say that temperature is a scalar field and that a scalar field is a mapping from space-time points (ambiguous!) into the reals, which seems to imply t.h at. t.emperature is a sort of mapping. One of t.he "is" clearly must mean "is represented by", but I can't. even say which because of the ambiguity of "field" as between physical entity and mathemat.ical representation.

'I

T. MAUDLIN

SUBSTANCES AND SPACE-TIME

space-time because of further structure defined on them. Also, . one might say that these space-time points are to represent events, m1rnmal regions of (physical) space-time.22 But this is a tendentious statement, since one of the questions raised by the argument we are cons1dermg 1s whether such points do or can represent events (cf. Norton 1987, p. 157). The first structure we invest the set of points with is lopological structure. This entails specifying which subsets of the set constitute open neighborhoods. This structure makes our set of points into_ a manifold (also called a topological or differentiable manifold). With such a structure we can now characterize continuous curves on the manifold and continuous mappings from one part of the manifold to another. We can lay down sets of co-ordinate curves, assigning to each point a unique n-tuple of real numbers. We can determine the dimensionality of the manifold by finding the minimum number of such co-ordinates needed to specify each point. We can define a denvative of curves. In the context of classical relativity theory we choose a 4dimensional manifold. Two points need to be made about the topological manifold. First, it does not yet have any structure which represents spatio-temporal properties and relations. No notion of distance between pomts 1s yet defined no distinction can be made between time-like and space-hke interva l s. There is nothing by reference to which lines can be characterized as straight or curved, and hence no grounds upon which to prefer one co-ordinatization of the manifold to another. There is nothing in the bare topological -manifold which would answer to the notion of spatio-temporal location. . . Second, the topological structure generally does not itself ascnbe different properties to different regions of the manifold. Although one may have regions which differ topologically from others because (l,oosely speaking) there are "holes" in the manifold, most neighbor_hoo(ls are topologically identical to one another. They mhent their md1v1dua­ tion, so to speak, from the fact that the pomts were md1v1duated ab inilio.23

To repeat: the topological manifold as such has no geometrical properties, and hence no mathematical resources to represent the spatio­ temporal structure of the world. Geometrical structure must be added, and this is done by means of two mathematical objects. Straight lines can be defined by adding an affine connection to the manifold, yielding an affine space. The affine connection ties together the tangent spaces at different points of the manifold, so that one has the notion of a vector pointing in the same direction (or pointing parallel to itself) as it travels along a curve. Spatio-temporal distance relations are represented by means of a metric defined at each point of the manifold, whose addition turns the manifold into a melric space. Since we want the straight lines also to be curves of shortest distance between points,24 we must adapt the affine connection and the metric to one another. If we further require that the connection be symmetric, then only one connection is compatible with a given metric, so that we can specify the entire geometrical structure of the manifold by specifying the metric alone. It is differences in the metric which distinguish between the mathematical space-times of Newtonian physics, of special relativity, and of general relativity. Newtonian space and time are represented by a degenerate metric which foliates the 4-dimensional manifold into a unique sequence of 3-dimensional spaces-at-a-time. The geometry of these spaces is the flat metric of Euclidean geometry." Special relativity employs the Minkowski metric, which creates a true 4dimensional space-time continuum. As in Newtonian theory, the metric is the same everywhere. Although the Minkowski space-time singles out no unique simultaneity relation between points, it does single out certain frames of reference or co-ordinatizations as special. These co-ordinatizations represent the inertial frames, in which the laws of physics take a particularly simple form, invariant between frames. Finally, the space-time of the general theory of relativity is represented by a manifold with a variable metric, whose geometry need not be Euclidean. The equations of the theory are generally covariant, which means that they are the same no matter what co-ordinate system is adopted. Both Newtonian and special relativistic theories can be given a generally covariant formulation, but they also admit of

446

(22) Events are often likened to physical occurrences infinitesimally localize? in space and time, like the explosion of an (infinitesimal) firecracker. However, th1 � .1s slightly misleading since there need not be a physical happening in a region of space-time in order that there be an event there. (23) One could reverse the cntological picture here, and posit the open neighbor­ hoods as the primitive individuals. One can then add a {transitive, assymetric) subset relation to the collection of neighborhoods and define points as ideal objects represented by certain infinite sequences of neighborhoods. Roughly, a point is a sequence or neighborhoods such that each is a subset of the preceding, and such that no neighbo:ho�d is a subset of all of those in the sequence. Two sequences represent the same point 1ff every open neighborhood in one sequence has a subset in the other. Ultimately �his �ay be a better ontological model for space-time, but because of certain countenntu1t1ve

447

features we won't pursue it. For example, the neighborhoods are only in a Pickwickian sense sets of points: one can define an "element of" relation (points are elements of the neighborhoods in the sequences representing them, a neighborhood has as elements all of the points in whose sequences it appears), but the "sets" here are ontologically prior to their "elements", reversing the usual picture. (24) Rigorously, we want them to be curves of extremal distance in a Riemannian metric. (25) Newtonian physics also presupposes a global condition that the space-time be topologically R4: infinite in all directions.

T. MAUDLIN

SUBSTANCES AND SPACE-TIME

particularly simple formulations in the inertial frames. In the general theory there are no such preferred co-ordinatizations. Now that we have a mathematical space-time, we can add more mathematical structure to represent the rest of the furniture of the world." We can add objects representing massive particles, electro­ magnetic fields, strong gauge fields, & c. Einstein's general relativistic field equations relate the distribution of matter and energy m the world to the space-time metric. The "bending ". of the metric d.ue to matter 1s gravitation. Since the general theory 1s formulated m a generally covariant manner, these obj ects and fields must be represented by mathematical objects whose characterization is independ ant of co­ ordinatization. Such mathematical objects are called tensors. In this context a physical theory is a set of differential equations which relate the various mathematical objects - tensors, metric, and connection - to one another. A model of a theory is a manifold with mathematical objects defined on it such that the equations of the theory are satisfied. Two models defined on the same manifold are different mathematical obj ects if they assign tensors differently at least one point. We are now in a position to undertake the problem which Earman and Norton raise for those who seek to regard space-time as a substance. First of all, take any model of the field equations. Let �s suppose that we have laid down some co-ordinates �uch that, m this solution, the co-ordinate curves are geodesics (straight Imes) m the manifold (given the metric of the solution). Further suppose that the equations say that material partides travel along geodesics. One bit of the model might look like this :

A diffeomorphism is a continuous one-to-one mapping from the manifold onto itself. It uniquely associates with each point in the manifold another point, in such a way that open neighborhoods are mapped onto open neighborhoods. Earman and Norton ask us to consider a hole diffeomorphism, that is, a diffeomorphism which is the identity map (maps each point onto itself) outside of some bounded region. Points are mapped to points other than themselves only within the hole: e.g.

448

material particle _..

Fig.

1.

-

co-ordinate curves

A material particle in space-time.

(26) Not everyone would continue in this way. The "geometrodynamics" of Wheeler proposes to represent material objects as kinks in the geometry of space­ time. We shall have cause to comment on this later.

J. A.

material particle ...

L -

I-'"

�

s

r- p·

co-ordinate curves

p

r+- a·

Q

N Fig.

I"

2.

449

R

R·

!.L

/

_.f/

- A diffeomorphic mapping.

One way to think. of this is as if the hole is made of rubber firmly attached at the edges and is pulled without tearing. Point P is pulled over to where point P' used to be, & c. In this way each point in the hole becomes mapped onto another point in the hole. So far we haven't done anything to the model, we have just defined a mapping from the manifold onto itself. But we can use this mapping to define a new distribution of geometrical objects on the manifold in the following way: pick up the metric and other physical tensors from each point and move them to their image points under the mapping.27 This induces a new (27) At this point I have radically oversimplified in order to avoid technical detail. The problem is that there is no way to compare tensors at different points in the manifold to see if they are "the same". You can't just pick up a tensor and drop it somewhere else. The trick is this: first express the tensors in co-ordinate form relative to the original co-ordinatization. Second, let the original co-ordinate curves generate a new set of coordinate curves, in which the new curves are the images of the old ones under the diffeomorphism. Of course, outside the hole the old co-ordinates map onto themselves. Finally, define the new tensors as those which have the same functional form relative to the new co-ordinates as the old ones had towards the olf co-ordinates. These new tensors are called the carry alongs of the old tensors under the diffeomorphism. See Earman's and No1'ton's papers for the mathematical details of the operation.

T. MAUDLIN

SUBSTANCES AND SPACE-TIME

assignment of tensors on the manifold. For examp le, our old model _ would yield a particle trajectory which would look hke this under the diffeomorphism:

indeterministic. For given any region, no matter how small, a complete specification of the physical state of the universe outside the region plus the field equations do not fix the physical state of the world within the region. Of course, so far we only have a conditional dilemma. If one regards the diffeomorphically generated mathematical models as representations of distinct physical possibilities, then radical indetermi­ nism follows. Not surprisingly, physicists do not so regard the models. They would consider that the two models represent the same physical state. But Earman and Norton argue that if one regards space­ lime as a substance one must accept the models as representations of distinct physical slates. Here, according to them, the substantivalist faces a dilemma: either he must accept radical indeterminism or abandon his substantivalism (cf. Earman & Norton ms. p. 15, Norton 1987 p. 180). Now the mathematics which Earman and Norton present, and which I have crudely sketched, is quite unexceptionable. However, the lynchpin of the arguments is not a piece of mathematics but a bit of metaphysics. It is here that we must focus our attention: What is it to regard space-time as a substance? How does such an ontology constrain one's interpretation of mathematical representations? For I do not believe that Earman and Norton establish that a substantivalist must be committed to regarding the models as representing distinct physical states. Indeed, I believe that they, lured on by the peculiar abstract ontology of mathematical representations, have fallen into the very problem that Aristotle uncovered in Metaphysics Z.3. But instead of attacking their arguments directly, I want to approach the problem a bit obliquely, by allowing the arch-substantivalist, Newton, to speak in his own defense. We must first note that there is nothing about the hole argument which makes it particularly appropriate only to the general theory of relativity. Models of the special theory, and of Newtonian dynamics, and even of Newtonian dynamics with absolute space are also 4dimensional manifolds with various geometrical objects defined on them." All of these theories can be given covariant formulations. So if the indeterminism problem arises, prima facie it arises for everybody. This is somewhat surprising since Newton's theory, especially with absolute space, has been regarded as a paradigm of both substantivalism and of determinism." Earman and Norton assert that

450

j

ma erial particle +

y1......

, / v s·

s

co-ordinate curves

Fig. 3. - The carried along particle trajectory.

Mathematically, this new model is a distinct object from the old. For example, whereas in the old model the material particle passed through point Q, in this model it doesn't (of course it does pass through the image of Q). But the vital result is the following: If the old model satisfied the field equations, so does the new one. This may at first blush look wrong. After all, the equation says that all material particles travel along straight lines, but now the trajectory seems , curved. But remember that in defining the model we picked up everything and moved it, including the metric, which determines which lines are straight. In the old model the co-ordinate curves were geodesics, but in the new model they are not (the images o f the co­ _ . _ ordinate curves would be geodesics of the new metric). i"ntuitively, it should be clear that the new model also satisfies the equations. For, as we noted, the topological manifold does not have any internal structure which would inherently distinguish one neighborhood from another or prefer one co-ordinatization to another. But the only difference between the two models is which particular neighborhoods the tensors occupy. The dilemma can now be posed. We have two distinct mathemati- . ea! models of the field equations which agree everywhere outside the hole but give different solutions inside the hole. I f one is obliged to _ _ regard each of these models as representing a distinct physical of affairs, then the laws which govern the world would be radically

1 983.

451

(28) A covariant presentation of the theories is given, for example, in Friedman

(29) John Earman has already revealed indeterminism of a rather different sort lurking around the edges of Newtonian theory, but even he regards the special theory of relativity as close to the ideal of deterministic theories. See Earman 1986 £or details.

T. MAUDLIN

SUBSTANCES AND SPACE-TIME

a metaphysics which forces us to be indeterminists must be unaccepta­ ble: determinism must at least have a fighing chance of being true, contingent on physical facts (Earman & Norton ms. p. 14, Earman ms. p. 20). Equally, it seems to me, an interpretation of the mathematical formalism which makes substantivalism imply indeterminism, irrespecti­ ve of the details of the physical theory in question, is unacceptable. I f not even Newtonian theory can b e interpreted substantivally without radical indeterminism, substantivalism is not being given a fighting chance.30 How, then, would Newton reply to the argument?

p. Newton the substantivalist would clearly brand such assertions as absurd.31 Earman and Norton's difficulty arises from asserting that the subslanlivalisl must regard space-lime as represented by the bare topological manifold. Since the metric tensor, as a mathematical object, is an accidental property of the manifold, they can then infer that the spatio­ temporal relations between physical regions must be accidental attribu­ tes of them. This results from mistakenly reading the ontological struclure of a mathematical representation into the objecl that ii represents. Since the mathematical structure is built up from bare particulars, all of whose properties and relations are accidental, one thereby arrives at a physical ontology of bare particulars all of whose physical features are accidental. That is, by denying that any physical attributes are essential to the entities which possess them one arrives at the prime matter of Z.3, with all of its attendant difficulties. The solution to this dilemma is just that which, I have argued, Aristotle saw. Not all predicates of a subject are accidental features of the subject. The essential features cannot be removed from the subject while leaving it behind. Similarly, moving the metric tensor from one point to another is a mathematical operation with no correlate in the realm of physical possibility. Physical space-time regions cannot exist without, and maintain no identity apart from, the particular spatio­ temporal relations which obtain between them. Since the ontological structure of the physical universe does not mirror the ontological structure of the mathematical object representing it, the mathematics must be supplemented with a metaphysical commentary. Earman and Norton themselves have done this in identifying space-time with the bare topological manifold and in insisting upon using a diffeomorphism, which preserves topological structure. For if one were to just adopt the ontology of the mathematics whole-heartedly, one would be forced to regard physical space-time points as entirely like the bare particulars from which the mathematical structure is constructed. But if one were to adopt this interpretation , one would not need to go through all of the technicalities of the diffeomorphism , for all one needs to do is permute the mathematical points to get a new mathematical object. That will generally alter the neighborhood structure, but that would just mean that the topological relations between physical space-time points are

452

As the order of the parts of time is immutable, so also is the order to the parts of space. Suppose those parts to moved out of their places, and they will be moved (i f the expression may be allowed} out of themselves. For times and spaces are, as it were, the places as well of themselves as of all other things. All things are placed in time as to order of succession; and in space as to order of situation. It is from their essence or nature that they are

places; and lhai the primary places of things should be movable, is absurd. (Newton 1686, p. 8, my italics.)

Newton insists that the parts of space and time have their spatio­ temporal relations essentially, and any talk of such parts changing their spatio-temporal properties is evidently confused. For if they do bear these relations essentially, by what (metaphysical) means could one identify a displaced place (one which had changed its spatio-temporal position relative to other places) as the very same place? To hark back to Aristotle, one might say that places are themselves in places, or that places are subjects to which the predicate "occupying such-and-such a relative position with respect to tliese other places" truly applies. But that predicate applies to a place essentially. However, Earman and Norton's interpretation of the hole diffeomorphism requires the substantivalist to reject Newton's claims. Under their interpretation, space-time points (physical, not mathematical, space-time points)• only possess their spatio-temporal relations accidentally. Points which ·were represented as being 2 meters apart in the original model, those very points, may be represented as being 8 meters apart in the diffeomorphi­ cally generated model. Points may be represented as changing their temporal sequence in different models: in one model p occurs in the absolute past of q while in another q occurs in the absolute past of

(30) Earman does give Newton a proposed response, along the lines that since his space-time structure is fixed and unaffected by matter distributions, he can r�gard all models as employing one and the same space-time (Earman ms. p. 19, cf. also Earman and Norton ms. p. 7). As will see below, the mutability of space-time, in Earman's sense, has no bearing on the question; and the immutability of places and times, in Newton's sense, is equally defensible in the context of the general theory of relativity.

453

·

(31) Howard Stein makes prescisely this claim about Newton in [Stein 1977], p. 394. Stein, however, is more generous than am I in countenancing locutions in other contexts which would imply that space-time does not have its spatio-temporal metrical properties essentially.

T. MAUDLIN

454

also accidental.32 Even Earman and Norton don't think the substanti­ valist must be committed to regarding permuted models as representing physically different states: they accept that the topological relations are essential to physical points. Physical space-time, according to their version of substantivalism, is represented by a set of mathematical points plus a topology. The question is: Why stop there? It is a most peculiar place to stop, for it commits the substantivalist to the belief that space-lime, and space-lime points, have no spalio­ lemporal properties essentially. If space-time is a substance in the universe, surely its appropriate mathematical correlate is an object which has enough mathema tical structure to represent spatio-temporal properties! The substantivalist's natural response to the hole dilemma is to insist that space-time is represented not by the bare marnfold but by the manifold plus metric, by the metric space. Earman and Norton do not just presuppose that space-time is to be represented by the manifold; they provide arguments to that . effect. Let us consider these arguments and some other conceivable objections to my substantivalist's position to see whether they establish their point.

Objection 1 : The metric is mathematically just like the tensors representing other physical fields . The physical field it represe�ts propogates, carries energy ,33 & c. Therefore the field must not be essential to space-time.

I answer that this is a non-sequitor. It only shows that, pace Newton, space-time has physical features which make it quite akin to other physical objects. All the more reason to regard it as a substance on a par with tables and chairs. One might try to turn this objection into a dilemma as follows:

Objection 2 : Since the metric is so similar to the other physical fields,34

how can one assert that

it

is essential to space-time and the others

(32) On the alternative mathematical ontology, in which neighborhoods are taken as primitives, this problem would arise in a different form: there the subset relation would be accidental, and neighborhoods could be permuted with respect to it. In some models p would be a subset of q. in others not. (33) Albeit in a somewhat peculiar way: the energy is represented by a pseudo-tensor rather than a tensor. See Earman and Norton ms. pp. 6-7 for arguments that the energy associated with the metric is just like all other energy. (34} There is possible escape hatch even to the objection that the metric looks too similar to the matter fields, although it is not a position I would defend. A substantivalist could argue that space-time is represented by neither the metric space nor by the bare manifold, but by an intermediate object. The affine space, the manifold plus affine connection, is just such an object. The connection is not a tensor, and the supposed problem of the gravitational energy pseudo-tensor vanishes since no such tensor is associated with the connection. The weaker structure of the affine connection still overcomes the hole dilemma, since the only affine transformation which leaves the area

SUBSTANCES AND SPACE-TIME

455

not? But . i� al� fields are essential, then space-time substantivalism becomes trivial: it only asserts that the whole universe is a substance.

I answer that despite the similarities, the metric has one distinguis­ . hmg property: it represents spatio-temporal features! Since space-time _ has its spat10-temporal features essentially (cf. Newton above), the . metric is essential to it and the matter fields not. Of course, if one could do physics without the metric, or show that the metric can be reduced to the matter fields, one could deny that space-time is a substance. As an aside, there are some peculiar mathematical features of the gravitational field equations which have been cited as evidence of a particularly intimate connection between the metric and space-time: The Ein�tein equations are non-linear. Actually, in this respect they are not so different from other fields, for while the electromagnetic field, the scalar field, etc., by themselves obey linear equations in a given space-time �hey form a non-linear system when their mutual interactions are take� into .accoun� . '_l'h e distin_ctive feature of the gravitational field is that it is _ self-tnler::zcitng: it 1s non-linear even in the absence of other fields. This is because ll defines the space-lime over which it propagates. (Hawking & Ellis 1973, p . 227,

�bj�ction 3 : " I f we do not classify such energy bearing structures as [a

my underline.)

grav1tat1onal] wave as contained within spacetime, then we do not see how we can consistently divide between container and contained. " (Earman & Norton ms. p . 7.)

I answer that the metaphor of container and contained is hopelessly obscure when applied to space-time. Space-time does not contain objects like a box contains objects (i.e. by surrounding them). It may contam objects hke the space within the box contains obj ects, but that IS not very enlightening . Two glosses of this objection are possible:

Objection 3a : �ravitational waves suggest that we cannot distinguish between the spatio-temporal and other aspects or properties of the universe.

I answer that in ordinary English, we have clear paradigms of spatio-temporal properties: distance, elapsed times, & c. We can easily

o.utside the hole unchanged (if it includes a Cauchy surface} is the identity map. The simplest way to .see this is. as follows: suppose a region of some thickness surrounding the Cauchy surface is to remain unchanged by the diffeomorphism (as will be the case with a finite h?le}. Since the . affine transformation (a transformation preserving the affine connect1on) .maps geodesics onto geodesics, it will map any geodesic passing through this _ onto itself. But any point within the hole will lie at the intersection of at least two region geodesics which pass through the Cauchy surface. Since these geodesics are mapped onto themselve� , e� ch point in the hole is mapped onto itself, and the map is the identity map : This still. allows the substantivalist to regard certain global transformations, such as moving e�eryth1n � 3 feet �o �he le.ft (in_ a flat space-time} or uniformly "stretching" space, as producing physically d1st1nct situations, but these transformations do not give rise to any problems about indeterminism.

456

T. MAUDLIN

SUBSTANCES AND SPACE-TIME

determine which mathematical structures in a theory represent informa­ tion about these properties, in this case the structure being the metric (as opposed to the electro-magnetic field tensor, & c.). If this is an energy-bearing structure, all the better for substantivalism, as per the response to objection 1 . I f we interpret "container" and " contained" as respectively "subject" and "predicate", we obtain:

which is the universe is as much correctly called "space-time" as it is " material body". Finally, the supersubstantivalist program is certain­ ly not trivial in the sense of being guaranteed to succeed: no one has yet made it work. Our best theories say that there is more to the universe than just space-time. In arguing that substantivalism about the general theory is inescapably subject to the hole dilemma in a way that Newtonian and special relativistic theory may not be, Earman introduces the notion that in the general theory space-time is mutable: Finally, assume space-time mutability in the sense that the object fields which characterize the structure of space-time are not given ab inilio but

Objection 3b : If the metric is essential to space-time, ?'e cannot distinguish between the predicate (metric) and that which has the predicate (space-time). But such a distinction is crucial to regarding space-time as a substance, and hence as a subject (bearer of fields).

I answer that regarding the metrical properties as essential to space­ time no more obliterates the distinction between subject and predicate than does regarding my humanity as essential to me. I am not identical with my humanity, I am the subject of which my humanity is predicated, I am the bearer of that property. Still, that does not make my humanity an accidental attribute, not does it make me into a bare particular which might have been, say, a chair. Space-time may be a bearer of the metric field but still bear it essentially. (The relevance of Metaphysics Z.3 is, I hope, vindicated here.)

Objection 4 : " Classifying the metric as part of the container spacetime leads to trivialization of the substantivalist view in unified field theories of the type developed by Einstein, in which all matter is represented b:f a generalized metric tensor. For there would no longer be anything contained in spacetime, so that the substantivalist view would in essence just assert the independent existence of the entire universe." (Earman & Norton ms., p. 7.)

I answer that belief that space-time is a substance does not become trivialized if it could be argued that space-time is the only substance (Is Spinoza's substantivalism about God trivial?). Indeed, programs such as Wheeler's "geometrodynamics" are correctly termed by Larry Sklar as "supersubstantivalist" (Sklar 1977, p. 221 ff.). The only sense in which unified field theories such as Einstein's could trivialize substanti­ valism is if the metric is artificially expanded and information about other structures just pasted on (in which case, the substantivalist response to Objection 2 would be undercut). Einstein's theory was arguably not of this sort: he attempted to represent electro-magnetism as an affect of the parallel structure of space-time, which is clearly a geomelrical property. Distinctions can usually be drawn 'between theories which try to show that physical effects flow from the metrical structure and those which artificially inflate the metric or the dimensionality of space-time to code up more information (cf. Weingard ms.). If in certain cases such a judgement were disputable, it would still be fair to say that if the paradigm geometrical properties of space­ time are ineliminably represented by the metric, then the substance

457

are regarded as dynamical objects on a par with the other fields. In any such theory substantivalism will be incompatible with Laplacian determi­ nism . . . (Earman ms. p. 1 8 , cf. also p. 19.)

Earman acknowledges the charge that the notion of mutability is "so vague as to be useless", but feels that paradigm cases are clear enough to deploy the concept. The question we must address is twofold: In what sense is the space-time of the general theory mutable? And what bearing does mutability have on substantivalism? It is unclear why Earman wants to designate the feature of being governed by local differential equations (rather than by global constraints) as "mutability". The only connection I can think of (which I would not attribute to Earman) is associated with a peculiar causal reading of the mathematics. The picture is that although the physical metric field is not given ab initio, the physical bare manifold is, and then the metric "propagates" along the manifold. This of course is nonsense: a model may represent objects propagating in space-time, and equations may state mathematical conditions relating the value of the metric tensor at some point to its value at points infinitesimally close, but the space-time metric does not propagate or "move". For the manifold and metric are not in time, time is in them. The notion of ab inilio which Earman invokes is obscure to me. A mathematician may, in explicating or constructing a mathematical object, begin by laying down some structure, but that does not mean that the object it represents is somehow metaphysically "laid down" first. "Ab initio" must here refer to the physical objects, not to the mathematical representation. But it cannot be given a temporal reading, and I don't know how to read it.35

(35) We might note in association with this that it is rather difficult, to clearly explicate the many causal locutions that are used about how the metric is generated or how it interacts with matter. Reconciling causal locutions with the God's eye view of the mathematician is deeply puzzling. This problem also arises in disputes about the coherency to talk about time travel into the past, and even raises its head in the passage from llawking and '-Ellis cited above.

i l l _____________________• •. _____________;,___________ _c _

11 I

458

T. MAUDLIN

' SUBSTANCES AND SPACE-TIME

It is true that according to the general theory of relativity the geometrical structure of space-time depends on the matter fields in the universe. In this sense, as opposed to Newtonian theory and the special theory, one cannot specify what the correct mathematical representac tion of the space-time structure is before one knows the disposition of objects. But if this is the appropriate sense of "ab inilio" then the manifold is also not given ab initio: matter distributions can also determine some of the topological structure of space-time, as when a black hole requires there to be a singularity. The straightforward reading of "mutability" suggests that space­ time as a whole, or regions of it, can alter . its geometrical properties. Clearly, space-time cannot do this in the sense that a persisting 3-dimensional object can, i.e., by having as a whole different shapes at different times. Different parts of space-time may have different geometrical properties, but that no more implies mutability than does the fact that different parts of the Hope diamond have different shapes. Hence, the only viable reading of "mutable" here is counterfactual: space-time is mutable in the sense that parts of it, or it as a whole, might have had geometrical properties other than the ones they do have. But such a counterfactual is not implied by the mathematical form of the field equations! The substantivalist would evidently deny that, strictly speaking, this very space-time could have had different spatio-temporal properties. Of course, had the matter fields been different, a space-time very similar to this one might have existed. It is quite open to the substantivalist to be a counterpart theorist about the counterfactual locutions that physicists are wont to use. There might have been other spacetimes, as would have been described by the various non-isometric models of the theory. Some of these are sufficiently like our space-time to provide counterparts to regions of actual space-time and so to support counterfactuals about what would have been the case if . . . But those counterparts are not these very spatiotemporal regions.36 In any case, counterfactuals about space-time, like all counterfactuals, are puzzling beasts, and it is not obvious that the non-substantivalist will fare any better. '

The only non-question-begging sense in which the space-time of the general theory of relativity is mutable, then, is the sense in which its geometrical properties are posited to depend upon the matter fields via the Einstein field equations, so that different parts of space-time may have different metrics and models of the theory may be non­ isometric. What bearing does this have on substantivalism? One possibility is:

(36) This is not to say that one must be a modal realist to be a substantivalist. Kripke might object that we can stipulate that we are counterfactually talking about this uery region (cf. [Kripke 1980) pp. 43-47), but he runs afoul of the very problem of essential properties. For he admits that not every such stipulated possibility is metaphysically possible; there is, for example, no possibility that Nixon might have been a chair (cf. ibid. pp. 1 10-116). Our substantivalist can adopt. Kripke's view of counterfactuals, but must insist that the geometrical properties of an individual space­ time be considered essential to it.

I answer that although the substantivalist must acknowledge the movement of every body 3 feet East as resulting in a new physical

·

459

Objection 5 : Substances are supposed to be capable of independent existence. You say that space-time has its geometrical structure essentially. But then it should be capable of independent existence apart from the matter fields. But if we remove the matter fields, we change the metric (via the field equations} and so destroy the space-time.

I answer that this objection rests on a confusion of physical and metaphysical necessity. If all American families were increased by one member, then the average American family would as a logical consequence also be changed. The average family cannot exist independently of the flesh-and-blood ones, and its properties are entirely determined by theirs. No such logical or necessary connection is evident in the case of the metric.37 The substantivalist can regard the field equations as conlingent truths, so that it is metaphysically possible for a particularly curved space-time to exist even if all of the matter in it were annihilated. Similarly, I cannot physically exist without there being surrounding air, but this does not debar me from being a substance.

Objection 6 : So far, you have only dwelt upon what the space-time substantivalist need not be comrr1itted to: he need not regard space-time as represented by the bare manifold, need not regard counterfactuals as indicating possible geometrical states of this very space-time, & c. Let us consider what the substantivalist must accept. Substarttivalists, whatever their stripe, "must all agree concerning an acid test of substantivalism, drawn from Leibniz. If everything in the world were reflected East to West (or better, translated 3 feet East}, retaining all the relations between bodies, would we have a different world? The substantivalist must answer yes since all the bodies of the world are now in different spatial locations, even though the relations between them are unchanged . " But "the diffeomorphism is the counterpart of Leibniz's replacement of all bodies in space in such a way that their relative relations are preserved. . .. In sum, substantivalists, whatever their precise flavor, will deny: Leibniz equivalen­ ce: Diffeomorphic models represent the same physical situation." (Earman & Norton ms., pp. 10-1 1 ; cf. also Earman ms. p. 17; Norton 1987 pp. 179-80.)

(37) There may be some such.non-evident connection, as the geometrodynamicist would cont.end, but this must be argued for Even for him, though, space-time is certainly capable of existing independently of anything else, it is just that there is nothing else for it to be independent from. ,

T T. MAUDLIN

SUBSTANCES AND SPACE-TIME

situation, it in no way follows that the diffeomorphism is the appropriate generalization of such a change. Rather, it would be the movement of the mailer fields which a contemporary Leibniz would have in mind, not moving the metric as well. The substantivalist does regard the mathematical object created by replacing all of the tensors save the metric tensor by their carry alongs under a diffeomorphism as representing a new physical situation. In general, of course, this new physical situation won't even satisfy the field equations. It is a special feature of Newtonian and special relativistic physics that moving all material bodies in a particular solution an equal amount will results in another solution. In general relativity there is no such automatic recipe for generating one solution on the metric space from another. Indeed, in general relativity the demand that one translate everything 3 feet East while retaining all the relations between bodies is not usually satisfiable. For example, if we move a physical triangle 3 feet East from a region of zero curvature to a region of positive curvature, the parts of the triangle cannot retain all of their geometrical relations to one another. Either the sides will no longer be straight or the interior angles will no longer sum to " · There is no reason to suppose that the Leibniz experiment generalizes to moving the geometrical structure of space-time as well as the material bodies, and good reason not to suppose so. First of all, if it did include the geometrical structure, it would not be necessary to specify that one should move everything an equal distance. Indeed, only in the case in which the metric is left unchanged does it make sense to talk unequivocally about "the distance" that an object has been moved. For suppose that under a diffeomorphism such as Earman and Norton propose one moves all of the geometrical structure from point p to point q. It will not generally be the case that the spatio-temporal distance between p and q remains invariant under the transformation! Either there is no univocal distance between the two points at all or else, if one insists upon measuring distances according to the metric from which the move is made, it will generally o.c cur that a move 3 feet East from p to q followed by a move from q 3 feet West will not bring you back to p. It is doubtful that Leibniz could have had such a possibility in mind. But even if he did, the substantivalist can rightfully point out that it begs to the question to say that all spatio­ temporal relations have been retained in such a move. For the spatio­ temporal relations between p and q, more generally between almost any pair of points in the hole, will have been changed. 11, as Earman and Norton insist, the substantivalist must regard the bare manifold as a substance, these he can respond that the diffeomorphism does not leave all spatio-temporal relations unchanged. 11, as I have maintained, he regards the metric space as the substance, then moving bodies around in

a given space-lime is a different mathematical operation than the one Earman and Norton present. Thus our substantivalist does pass the acid test - appropriately construed. In Newtonian dynamics and special relativity he would accept the movement of all matter fields or particles 3 feet East as resulting in a new physical situation, one which still satisfies the field equations. In general relativity, he would accept the moving of matter fields globally also as creating a new physical situation. If the metric were sufficiently isotropic and homogeneous, a global transformation similar to moving everything 3 feet East might be definiable and even lead to a new solution to the field equations. But accepting such global transformations as resulting in distinct physical states raises no spectre of indeterminism, for in such a case the Cauchy surface must be changed too. On the other hand, moving the matter fields around according to a hole diffeomorphism will not lead to a solution to the field equations unless it is the identity map. Accepting Leibniz's "acid test" does not need to imply indeterminism.

460

·

461

Objection 7 : Your space-time substantivalist has escaped only tempo­ rarily by a semantical trick. Suppose we grant that the term "space­ time" does refer to an entity best represented by the metric space rather than the topological manifold. Still, that entity does have topological features. If we consider it qua only its topological features, we are considering something, something which by definition does not have metrical features essentially. I will give you the term: space-time: let us call this new entity "topo-space". Now if space-time is a substance, so is topo-space, and we can just repeat the argument for indeterminism.

I answer that it is true that one can inquire into the topology of a metric space, and determine what features it may have in virtue of that topology. Whatever conclusions one draws will equally apply to other metric spaces with the same topology even though their metrics may differ. The details of the metric may then seem to be irrelevant to the space-qua-topological. But this is not true if the entity only has its topology in virtue of its metric. For example, one may argue that the neighborhood structure is derived from the metrical features: sets of {physical) space-time points only form neighborhoods because of their spatio-temporal proximity relations. Thus the topology flows from the metric rather than the metric being imposed on the topological space. If one thinks of metric and manifold as form and matter, this is the ultimate case of the essential form making the matter what it is. Nor is this response purely a philosopher's fantasy. Einstein expresses precisely such a view in a passage which Norton cites: If we imagine the gravitational field, i.e. the functions 9ik, to be removed,

there does not remain a space of the type (I} [Minkowski space-time], but absolutely nothing, and also no "topological space". For the functions 9ik describe not only the field, but at the same time also the topological and metrical str1;1ctural properties of the manifold. (Einstein 1917, p. 155.)

I. i..--- 1 ·1

462

T. MAUDLIN

SUBSTANCES AND SPACE-TIME

But whereas Norton, believing that the substantivalist must hold his view towards the topological space, takes this as evidence that Einstein rejected substantivalism, I contend that this is just the position the substantivalist must hold (see appendix). None of this constitutes a positive argument for regarding space­ time as a substance. Like statements about the average American family, space-time locutions might yet turn out to be misleading ways of talking about material objects and their relational propertie_s. But it does not follow from the fact that the space-time metric physically depends on material fields (via the field equations) that it can be logically reduced lo relations between those objects. Nor, as we have seen, need the substantivalist accept radical indeterminacy; he need only insist that mathematically the metric space, not the topological manifold, represents space-time. One must resist the temptation to read the ontological structure of mathematical objects into the objects that they represent. And, as Aristotle saw, one must be extremely sensitive to the relationship between substances and their essential properties if one is to avoid the absurdity of bare physical particulars, or bare physical manifolds. With Aristotle's help, we have been able to turn back one set of objections to the view that space-time is capable of existence independent of material bodies. A close analysis of the positive arguments for that view must await another time and place.38

Although I agree with the general thrust of Norton's main conclusion, I do not believe that the texts will al bear the interpretation he gives them. First of all, I do not believe that all four passages are versions of the same argument. The first three, we are told, are essentially identical, and I shall so treat them {cf. ibid. p. 163). But the fourth has a different, more complex structure. It is only the fourth which obviously employs an active redistribution of the metric in the manifold, although Norton is correct in pointing out that the first three {hereafter known as the " first argument") turn upon more than the possibility of different co-ordinate expressions for a tensor at a given point. Let us begin with the first argument, and come to the second argument in turn. I reproduce Norton's translation of the first argument: If the reference system is chosen quite arbitrarily, then in general the gmn cannot be completely determined by the Tsn [i.e. the stress-energy tensor density]. For, think of the Tsn and gmn as given everywhere, and let all Tsn vanish in a region Cl> of four dimensional space. I can now introduce a

Appendix: Einstein and the Hole Argument

In [Norton 1987], John Norton argues that Einstein had discovered the hole argument by 1913. Norton contends that the argument was presented in four places {Einstein & Grossman 1913, Einstein & Grossman 1914, Einstein 1914a, and Einstein 191 4b), but that it has since been misunderstood. According to him, Einstein's argument has been misread as employing only a passive re-co-ordinatization of the manifold and as incorrectly trading on the fact that the same tensor is expressed as having . different components in different co-ordinate systems {Norton 1987, p. 164). One ought, he says, to read the transformation actively, as a diffeomorphism, rather than merely as a re-co-ordinatization. Thus, the hole argument was born over 70 years ago.

(38) I address these issues in "Buckets of Water and Waves of Space: Why Space­ time Probably is a Substance", which was originally intended as, and may ultimately be, the third part of this paper. At this point, the final disposition of all of the parts as independent or conjoined awaits further clarification.

463

new reference system, which coincides completely with the original outside

Cl>, but is different to it inside Cl> (without violation of continuity). One

now relates everything to this new reference system, in which matter is represented by T'sn and the gravitational field by g'mn· Then it is certainly true that T'sn = Tsn everywhere, but against this the equ_ations g'mn = gmn will definitely not all be satisfied within Cl>. * The assertion follows from this. If one wants a complete determination of the gmn (gravitational field) by the Tsn (matter) to be possible, then this can only be achieved by a limitation on the choice of reference systems. * The equations are to understood in such a way that each of the independent variables x'n on the left hand side are to be given the same numerical values as the variables x0 on the right hand side. (Ibid., pp. 163-4, Einstein's italics and footnote.)

Norton goes on to claim that the only way to understand the footnote is to see Einstein as invoking a diffeomorphism rather than just a passive co-ordinate transformation.39 But a much more straightfor­ ward reading is possible.

(39) I do not want to reproduce the algorithm which Norton takes this footnotes to indicate, because it is rather complex and, to me at least, not entirely clear. As far as I can tell, Norton is seeking an algorithm for comparing gmn(hp) [the original metric at hp expressed in terms of the original co-ordinate system] with h *gmn(hp) [the carried along metric at hp expressed in the original co-ordinate system]. If so, his algorithm should apply to gmn(P) the transformation for x'n-+ xn at hp, rather than the transformation xn-+x'n at p. Norton algorithm will probably compare gmn(hp) with gm'n'(p) (I say "probably" because he does not tell us in what co-ordinate system we are to express the original metric at ' p).

111.. --------------------llll.. llllii ______________________ l

I

lJ

�'

465

T. MAUDLIN

SUBSTANCES AND SPACE-TIME

The argument that Einstein gives does employ a hole, but the first and last sentences make quite clear that the question at hand is not whether the entire state of the universe the outside the hole determines the state inside. Rather, the question is whether the stress-energy tensor density Trn (representing matter) defined everywhere determines the metric gmn everywhere. Einstein thinks that it must, probably as a sort of application of Mach's principle that intertial structure is determined entirely by the disposition of masses. The e roposed criterion for an acceptable theory is the following: if you lay down a co­ ordinate system and specify T"' everywhere as a function of those co­ ordinates, then the gmn expressed as a function of those co-ordinates should be fixed by the laws of the theory. Einstein shows by a reduclio ad absurdem that this requirement cannot be fulfilled unless one restricts the reference systems which are allowed. The reduclio runs as follows. Consider a distribution of matter T such that T 0 within a region . Consider two co-ordinatizations of the space-time x" and x'" which agree outside the hole but diverge within it. Then, since a zero tensor is expressed as having components equal to zero in all co-ordinatizations, the expression of T as a function of the x" will be funclionally identical to the expression of T in terms of the x'n . That is, Tmn = T'mn everywhere, and "everywhere" can be taken in two senses. It is true that the co-ordinate components of T at any given point p in terms of the x co-ordinates are equal to the co-ordinate components of T in terms of x' co-ordinates for that point p [(V p)(Tmn(P) = T'mn(p)]. But it is equally true that the co­ ordinate components o f T in terms of the x" al a point with co-ordinate values < N 1 , N,, N , , N 4 > in the x co-ordinate system is equal to the co-ordinate components of T in terms of the x'" al the point with the same numerical co-ordinate values in the x' co-ordinate system [(V N I)(V N ,)(V N,)(V N ,)(Tmn( < N , , N,, N , , N, > x) T'mn( < N I , N , , Na, N 4 > x.J].40 The points referred to on either side of this equation are points which have a particular co-ordinate expression in each reference system, and since the systems diverge within the hole they will, in ' general, be different points of the manifold. It is this latte r i dentity which, as his footnote illustrates. Einstein had in mind. It is this latter equation which correctly expresses the fact that the stress-energy tensor density has the very same functional form when expressed in

terms of the x" as it has when expressed in terms of the x'". But now an application of the Machian principle cited above yields that if the theory appropriately allows the T mn to determine the gmn, then the gravitatio­ nal metric must also have the same functional form relative to each co­ ordinatization, gmn = g'mn · This statement again is made .for points with equal co-ordinate values, i.e., for different points in the manifold. But this clearly cannot be the case, since different points within the hole may have, e.g., metrics with different (scalar) curvatures: the gravitational field cannot generally take the same functional form when expressed in terms of the two co-ordinate systems. Einstein's footnote makes clear that he is comparing the gmn and the g'mn at different points. Since the points are to have the same numerical co-ordinate values but are given in two different reference frames, they must in general be different points. In a scientific world in which changes of reference frame were common coin and diffeomor­ phisms not, the equations could not have been read otherwise. The argument is crisp and accurate, and it does indeed demonstrate that generally co-variant field equations cannot satisfy the Machian criterion stated above. This failure does not simply arise from gauge invariance (pace Pais 1 982, pp. 221 -222), nor is the point identical with the hole dilemma of Earman and Norton. It just is a fact that fixing T does not fix g in general relativity. There are, for example, solutions of the vacuum field equations with and without gravitational waves, and hence space-times with T 0 everywhere which are not isometries. The criterion that Einstein states at the beginning of the passage cannot be satisfied by the general theory. The first argument, then, makes perfect sense in light of the Machian principle enunciated in the first sentence. No diffeomorphism need be invoked in understanding his footnote, only a re-co­ ordinatization. Further, we later find Einstein holding to such a criterion and criticizing the equations of 1915 in its light. In 1917 Einstein finds fault with his own earlier analysis of planetary motion because it implies that "the intertia [of a body] is influenced by matter (at finite distances) but not determined by it. If only a single mass point existed it would have inertia ... [but] in a consistent relativity theory there cannot be inertia relative to 'space' but only inertia of masses relative to each other" (Einstein's italics. Cited in Pais 1983 p. 286). And in 1918 he states as a fundamental principle that "The G­ field is without remainder ('restlos') determined by the masses of bodies" (cited in Pais p. 287). With the passage of time Einstein came to see that this Machian demand could not be met, and so he eventually abandoned it (cf. ibid. pp. 281-288). Einstein's first argument does not employ an active transformation like Norton's diffeomorphism . Furthermore, it is not as general as his

464

=

=

(40) In these equations, I have used terminology consistent with Einstein's, as I interpret him. "T'mn" refers to the co-ordinate expression of T as a function of the x'n. A 1nore perspicuous terminology would be T which would make evident that the prime indicates a change of co-ordinates, not a change in the tensor itself. In Einstein's second argu1nent he had to introduce some such more perspicuous notation, indicative of a change in the mathernatical operation being performed. m'n',

=

!

Ii ·

I. ;�

, SUBSTANCES AND SPACE-TIME

since it requires that the stress-energy tensor density be identically zero within the hole, so that its co-ordinate representation under the re-co­ ordinatization remains the same. His second argument, however, clearly does employ the notion of an active transformation. The extra · complexity of this argument is reflected in the additional technical machinery he is required to use: whereas in the first argument he could unambiguously refer to gmn and g'mn, since one metric was at issue, albeit described in two reference frames, in the second argument he needs " G(x)" to refer to the original metric expressed in the originar frame, " G'(x')" to refer to the original metric as expressed in the new co­ ordinate system,41 and G'(x) to refer to the metric with the same functional form as G'(x'), but with the arguments now being the co­ ordinates of the first system (cf. Norton 1987, p. 167). Thus although I don't think the texts support Norton's active reading of the first argument, there clearly is an active transformation in the second. It is not exactly the hole argument of Norton and Earman, but it is similar. What did Einstein make of this argument? Originally, he inferred that general co-variance must be sacrificed. But he later came to accept, even embrace, that requirement. What had to be given up to remove the fangs from the hole argument? Norton asserts that at first Einstein could only resolve the dilemma by adopting non-realism about space-time, by which he means the view that the English term "space-time" has "no referent in the physical world" (ibid. p . 177). Arthur Fine evidently takes a similar view (cf. Fine 1984, pp. 91-2). Norton does think this inference of Einstein's to be unwarranted (Norton 1 987, pp. 176-83), but accepts it as a temporary error on the way to a weakened "anti-substantivalist" view of space­ time (which we shall take up presently). Both Norton and Fine can cite some passages from 19 15-1916 in which Einstein appears to take up this uncompromising position: "That this requirement of general co­ variance, which takes away from space and time the last remnant of physical objectivity, is a natural one, will be seen from the following reflexion . . . " (Einstein 1916, p. 1 1 7); "The essential thing is: as long as the drawing paper, i.e. 'space', has no reality, then there is no difference at all between the two figures . . . " (cited in Norton 1987, p. 174). If we take these quotations in isolation, and ascribe to Einstein some understanding of referential semantics, Norton's claim appears to follow.

But a close examinati on of the context reveals that the non-realist interpretation of these passages does not capture Einstein's meaning. For Einstein consistently interchanges assertions of the form "space and time are not physically real" with assertions of the form "spatial and temporal co-ordinates have no physical significance ", _and it is clearly the taller claim which concerns him. Consider the begmnmg of the section on general co-variance from [Einstein 1916]:

(41) It is notable how Einstein uses his primes here, for it reflects the way he thought about the problem. To someone schooled in co-ordinate free presentations, G(x') would be the natural way to denote the original metric as expressed in the new co-ordinate system, for the metric (G) is the same, only the reference frame has changed. But what was important to Einstein was the funclional form of the metric expressed in terms of the co-ordinatization. Since that functional form changes under re-co-ordinatization, the same metric becomes G'(x') rather than G(x').

. ,'1

467

T. MAUDLIN

466

In classical mechanics, as well as in the special theory of relativity, the co­ ordinates of space and time have a direct physical meaning. To say that a . . point-event has the X1 co-ordinate :.z:1 means . t�at the proJ.ect1on of the point-event on the axis of X1 , determined by. r1g1d rods and 1n accord�nce . with the rules of Euclidean geometry, is obtained by measuring off a given rod (the unit of length) x1 times from the origin of co-ordina �es along the t, axis of X1 . To say that a point-event has the x4 co-ordinate X4 means that a standard clock, made to measure time in a definite unit period, and which is stationary relative to the co-ordinates and practically coincident in space with the point-event, will have measured off X4 t periods at the occurence of the event. This view of space and time has always been in the minds of physicists, even if as a rule, they have been unconscious of it . . . ' (P. 1 15, my italics.) =

=

This "view of space and time" clearly has not to do with the ontological status of space-time but with the operational significance of spatio-temporal co-ordinates.42 In Newtonian physics and in the special theory of relativity, once you associate an inertial frame with a physical body as it's origin and fix some co-ordinate axes and a zero time the assertion that an event will occur at, say, < 1 , 3, 0, 7 > has phys ical significance. It tells you about the behavior of clocks and rigid _ rods connecting the body at the origin with the event. And co-ordmate differences between points have a similar meaning. Einstein's point in denying physical reality to space and time in a generally co-variant theory is that no such operations are associated with co-ordinate values in such a theory. The most exhaustive informat10n about the or1gm and axes of a reference frame in a generally co-variant theory will impart no information about how to physically locate particular co­ ordinatized points. Co-ordinate differences also obviously will have no general physical significance. . Perhaps the physicist who Einstein speaks of as unconsc10usly holding this view of co-ordinatizations was himself several years earlier. That would in part explain why, in the first argument, he thought it significant that points with the same co-ordinate value would have different metrics. Since the axes of the reference frames could die

(42) Of course, given a sufficiently positivist outlook, questions of ontology or reality may simply reduce to questions of operational significance, and the question concerning us, whether space-time is a substance or not, cannot even be formulated.

468

T. MAUDLIN

entirely outside of the hole, if some operational significance akin to that in the special theory could be ascribed to the co-ordinates here, general covariance would have to be curtailed. Once one shakes off the notion that co-ordinates have any operational meaning, a more complicated argument, such as the second one, is needed. Einstein's concern with the operational significance of co-ordinates rather than the ontological status of space-time is also evident in the letter to Ehrenfest from which the second of the above "non-realist" quotations was taken. Earlier in the letter he writes "Your difficulty has its root in the fact that you instinctively treat the reference system as something 'real' " (Norton 1987, p. 1 73). But the "imaginary" character of the reference system, in this sense, need not impugn the reality of the object to which the system is attached. Differences in Dewey decimal numbers signify no real physical property, but that need not make us non-realists about the contents of the public library. If Einstein ever thought in terms of substances, non-substances, and non­ existent entities, these passages would not commit him to non-realism about space-time itself. Co-ordinate values in Newtonian theory and in the special theory can have a physical significance because what is co-ordinatized is the metric space, and because only a limited number of reference systems are allowed. These are two quite different points. Certain particular co­ ordinatizations of these metric spaces yield especially simple co-ordinate expressions for the physical laws because these space-times exhibit a perfect isotropy and homogeneity. Curvalinear co-ordinate systems can be put down on these spaces, and the laws expressed in terms of them, but then the physical significance of the co-ordinate values becomes much more complicated, and no general significance can be ascribed to co-ordinate differences. Still, given any co-ordinate system laid down on a metric space, whether in special or general relativity, physical statements of the form "a clock moving from point p to point q along path I will register 5 ticks" can be made. Once one moves to co-ordinates on the bare manifold, however, no physical significance, no mailer how complex, can be associaled with the reference system. Points in the bare manifold are no particular distance apart, stand in no causal relations, are not intrincisally connected by . any physical process. Thus, if, in one way or another, physical processes are used to define reference systems, we find that one can only meaningfully co-ordinatize space-lime as a metric space. This is one way of stating the hole dilemma. Given Einstein's equation of the reality of space-time with the physical significance of frames of reference, it is not surprising that he should come to regard space-time as real only if it is invested with a metric. This is the position that Norton dubs "anti-realism'', and is best captured by the following quotation :

SUBSTANCES AND SPACE-TIME

469

There can be no space [space-time] nor any pa�t of SJ? ace wit�?ut gravitational potentials; for these confer upon space its me�r1cal qualities, without which it cannot be imagined at all. The existence of the gravitational field is inseparably bound up with the existence of space. (Cited in Norton 1987, p. 181.)

I f this is taken to be the core of Einstein's position, then our metaphysical battles reduce to a mere squabble over terminology. Norton calls this position anti-substantival, but I hope to have shown that it is just what any red-blooded substantivalist, from Newton on, would insist upon! Space-time must be regarded as essentially a metric space, not a topological space which has its geometrical properties only accidentally. The real question is whether such a metrical space with intrinsic inertial structure must be considered to be capable of existing independently of material bodies (represented by the stress-energy tensor), or whether is can be analyzed away as a . fafon de parler about the relations that obtain between bodies. This 1s the issue which divided Newton on the one hand from Leibniz and Mach on the other, the issue which impelled Einstein to his original insistence that the stress-energy tensor completely determine the gravitational field. Einstein grew away from this demand, and in so doing abandoned the relationalist camp. I hope to have shown that Einstein was quite correct in his pronouncement, a year before his death, that "one should no longer speak of Mach's principle at all" (cited in Pais 1982, p. 288). If one accepts the general theory of relativity as giving a literally correct picture of the physical world, one must stand with the Newtonian tradition and regard space-time as a substance.

REFERENCES

ARISTOTLE [1970] Aristotle's Physics I, II, translated with notes by W. Charleton, Clarnedon, Oxford. [1975] Posterior Analytics, translated with commentary and notes by Jonathan Barnes, Clarendon, Oxford. DANCY, R. [1978] "On Some of Aristotle's Second Thoughts About Substances: Matter". Philosophical Review, LXXXVII, No 3 (July 1978) pp. 372-413. EARMAN, J. (1986] A Primer on Determinism. Western Ontario Deries in Philosophy of Science, vol. 32, D. Reidel. [ms.] "Why Space is Not a Substance". and NORTON, J. [ms.] "What Price Spacetime Substantivalism? The Hole Story", forthcoming in the British Journal for the Philosophy of Science. EINSTEIN, A. [1914a] "Prinzipielles zur verallgemeinerten Relativitaetstheorie". Physi­ calische Zeitschrift 15, pp. 176-80. [1914b] "Die formale Grundlage der allgemeinen Relativitaetstheorie". Preuss. Akad. der Wiss., Sitz., pp. 1030-1085.

B10LOGJE. LOGIQUE ET M£TAPHYSIQUE

T. MAUDLIN

470

chez AiusToTE

[1916] "The Foundation of the General Theory of Relativity" in The Principle of Relativity, Dover, 1952. [1917] Relativity: The Special and General Theory. Methuen, London 1977. [1956] The Meaning of Relativity, fifth edition, Princeton. and CARTAN, E. [1979] Lelters on Absolute Parallelism 1929-1932. Princeton. and GROSSMANN, M. [1913] "Entwurf einer vera\lgemeinerten Relativitaetstheorie und einer Theorie der Gravitation". Zeilschrifl fuer 1'1aihemalik und Physik 62, pp. 225-261 . and GROSSMANN, M . [1914] "Kovarianzeigenschaften der Feldgleichungen der

auf die verallgemeinerte Relativitaetstheorie gegruendeten Gravitatiof!_slheorie".

Zeilschrift fuer i\11.alhemaiilr und Physik 63, pp. 215-225. FINE, A. [1984] "The Natural Ontological Attitude" in J . Leplin Realism. University of California Press, Berkeley, pp. 83-107. FREIDMAN, 1\.-1. [1983] Foundations of Space-Time Theories. Princeton.

(ed.),

Scientific

G11.1., M. L. [ms.] "Aristotle on the Unity of Definable Substances", forthcoming. llAWKING, S. and ELLIS, G. F. R. [1973] The Large-Scale Structure of SpaceTime. Cambridge University Press. KRIPKE, S. [1980] Naming and Necessity, Harvard. NEWTON, I. [1986] Principia, trans. by A. Motte, revised by F. Cajori, University of California Press, Berkeley, 1 966. NORTON, J . [1987] "Einstein, the Hole Argument, and the Objectivity of Space" in J . Forge (ed.) Measurement, Realism, and Objectivity, D. Reidel, pp. 1 53-188. PAIS, A. [1982] Subtle is the Lord.. The Science and Life of Albert Einstein. Clarendon, Oxford. RoBINSON, l-1. M. [ 1974] "Prime Matter in Aristotle". Phronesis 19 ( 1974) pp. 168-188. SKLAR, L. [1977] Space, Time, and Spacelirne. University of California Press, Berkeley. STEIN, H . [1977] "On Space-Time and Ontology: Extract from a Letter to Adolf Grunbaum" in Foundations of Space-Time Theories. Minnesota Studies in the Philosophy of Science, vol. V I I I , ed. J. Earman, C. Glymour, and J . Stachel, Minnesota Press. h WEINGARD, R. [ms.] "Early Unified Field T eories".

SCminaire CNRS-N.S.F.,

1987

Editions du CNR.S, Paris, 1990.

MAN AND COSMOS IN ARISTOTLE: Metaphysics A and the Biological Works1 ANTHONY PREUS

INTRODUCTION

·

In what sense is Aristotle's biology continuous with the twentieth century science of biology, and in what sense is it discontinuous, a separate enterprise? Critics of Aristotle's biological works are fond of citing his various "errors" : the notions that the heart has just three chambers,2 that women have fewer teeth3 and fewer sutures in their skulls,' for example. But all such critiques start by assuming that Aristotle was engaged in an enterprise not dissimilar from modern natur1l history. In fact Aristotle reports so many accurate observations which could be part of a modern natural history, that his "errors," from this point of view, would not lead us to suspect that his enterprise is something quite different and alien from anything which may be found in modern science. Another group of supposed "errors" includes his claim that the seat of the soul is in the heart (rather than in the brain), that it is the blood (1) This paper was presented at the conference on Aristotle's Metaphysics and Biology at lie de l'Oleron in June/July 1987. Part of my preparation was aimed at commenting on a potential paper by Michel Roussel; I have profitted by reading some of his essays: "Physique et biologic dans la Generation des Animaux d'Aristotle, "Revue des Eludes Grecques XCI I I (1980) 42-71"; E ther et chaleur dans l'embryologie aristotelicienne, influences archalques", in Melanges Lebel, pp. 157-160, Quebec 1980, and "La semence de feu" in Melanges Gareau, pp. 57-64. I thank several participants in the conference for calling attention to some of my errors; Allan Gotthelf deserves special thanks, though I have no doubt still not met all of his objections. (2) PA 1 !1.4, 666b21, e.g. (3) HA 11.3, 50I b20 . (4) PA II.?, 653bl.

A. PREUS

472

vessels which carry information from the sense organs to the heart and back to the organs of local movement,' apparently by means of connate pneuma,6 that the female provides the matter and the male the form, in sexual generation,' and that male and female embryos develop in the uterus at different rates.8 In these cases, it is pretty obvious that the claims are "theory-laden", and that Aristotle's statements indicate the consequences of a theoretical stance which is radically but recognizably different from our own. Those who believe in "scientific revolutions" and "epistemological breaks" between Aristotle and us can- very well point to these passages, these theoretical trends, as evidence supporting their hypothesis that Aristotle's "science" is fundamentally different from our own, perhaps in ways that do not automatically entail "progress" from Aristotle's day to ours. Still, we can compare the theoretical structure of Aristotle's explanations and in many cases find that, according to the philosophical canons of our science at least, Aristotle's explanations appear to "fail" in comparison with those offered by twentieth century investigators. There are some passages which are much more "alien" to most of us as we read the biological works; we cannot read these passages as either ,, "successes" or "failures on our terms; their author is playing with an entirely different deck than we are. Some passages which strike me, at least, that way are the "cosmological" passages; the Generation of A nimals has several especially striking examples of such passages. It is impossible, I think, to interpret these passages from a standpoint taken within the biological texts alone; one must look at them from the wider perspective of Aristotle's general theory of the relationship between life, especially human life, and the cosmos in general. We find in Aristotle's bio-cosmology (if I may call it that) an apparent anthropocentrism which has become almost incomprehensible to us, a theory of ",heat" which we are tempted to call "mystical" (when we really mean "mysterious"), and an assumption of the relationship between mind and the universe which survives for us only as a dimly recognized religious dogma, not as anything which we can accept as cognitivel y significant. We may classify the passages in question in a variety of ways; for my present purposes I will discuss them according to the following schema: I . "Orientation:" Human erect stature m alignment with the cosmos.

1

MAN AND COSMOS IN ARISTOTLE

473

2. "Pneuma and ailher" (especially GA 1 1 .3). 3. Biological and Cosmic Periods (especially GA IV. 1 0). Valuable clues to the understanding of these passages in the biological books may be gained by reading them in the light of Metaphysics A; I begin by calling attention to two places in that book: Our forefathers in the most remote ages have handed down to us their posterity a tradition, in the form of a myth, that these [celestial] entities ar: gods a!1d �he divine encloses the whole of nature... We must regard th1s as an 1nsp1red utterance, and reflect that, while probably each art and _ science �a � often been developed as far as possible and has again perished, these op1n1ons have been preserved like relics until the present.9 We must consider also in which of two ways the nature of the universe con tains the good or the highest good, whether as something separate and . by itself, or as the order of the parts. Probably in both ways, as an army does. For the good is found both in the order and in the leader1 and more i� the latter; fo � he does not depend on the order but it d epends on him. And all things are ordered together somehow, but not all alike, both fishes and fowls and plants; and the world is not such that one thing has nothing to do with another, but they are connected. For all are ordered together as to one end .10

I take these passages as summaries and indications of the Aristotle's cosmological theories as contained not only in Metaphysics A but also in the Physics, de Caelo, and Meleorologica.

I) "ORIENTATION" The idea of cosmic directions (up and down, right and left, front and back) is more prominent in the Physics (especially IV.I) and de Caelo than in Metaphysics A , but the results of those inquiries seem presupposed by the argument in A. It may be too that the idea of cosmic directions is a relatively "primitive" element in Aristotle's thought; at any rate passages referring to the notion of cosmic ori:ntation are the most pervasive sort of "cosmic" passages in _ Anstotle's bwlogy. They occur in all of the biological works, as well as elsewhere. a) In Lhe Hislory of Animals 1 . 1 5 the upright position of man is simply stated to be "natural" with respect to the arrangement of the universe: In � an, above all ? ther � nimals, the _up per and lower parts are arranged 1n accordance with their natural pos1t1ons; for in him, upper and

(5) PA Ill .4, 666a1 1 et al. (6) MA 10, 703a12. (7) GA 1.20, 729a!O el passim. (8) HA VIl.3, 583b3.

(9) Metaph. A. 8, 1074bl-4, 9-12, Ross/Barnes tr., changing "subtance" to "entity". (10) Melaph. A. 10, 1075bl 1-19, Hoss/Barnes tr. The continuation of the passage is equally interesting.

474

A. PREUS lower are the same as in the universe as a whole. In li� e manner the par�s in front, behind, right and left, are in acco :d � nce. with nature.- But !n regard to other animals, in some cases these d1s�1nct1ons do no exist, ? nd 1n others they do so, but in a vague way. For u �stance, the head w1 th . all _ . has animals is up and above in respect to their bodies; but man alone , as _ been said, has, in maturity, this part uppermost in respect to the universe.11

b) Elsewhere, several explanations are given for the " naturalness" of man's erect posture. One such explanation, fairly frequent m the biological books, is that "up" and "food intake" have a natorally close relationship, and that "down" and " residue output" also have a . naturally close relationship. This point is made particularly cleady m IA 4, where up and down, front and back, right and left are said to _ define living beings. Up and down are universal, charactensmg plants as well: They are distinguished fun.ctionally, not ju �t b)'. position in relation to . earth and sky - for there from where the d1str1buti � n o � food and growt_h starts in each sort of thing, is up; the part at which it fin? l�y ends, �s down. The one is the origin, the other the finish; up is the origin. Yet it might seem that down is more special in plants; for in them up and down do not have the same position as in animals. They do � ot have the same relation to the universe, but they have the same function. For the ro?ts are the "up" in plants; from them the food is distributed to the growing parts, and they get it with them as animals do with mouths. 12

The passage continues to discuss the funcbons of front and back, _ right and left in animals, and makes clear that m huma n bemgs, up is _ more cle arly more clearly up, front is more clearly front, and right 1s _ right, than in any other animal. Now of course one might easily claim _ _ that the import of this whole line of d1scuss10n 1s simply to assert that

(11) HA I.15, 494a27-b1 , Thompson/Rarnes tr. In de Caelo I I.2, 28�b5 ff, �:istotle _ considers the Pythagorean hypothesis that right and left have a cosmic s1gn1f1cance; referring to his discussion of the movements of animals (clearly a referen�e to 11- 4-5, part of which we quote next in the text), Aristotle first distinguishes the coordinates in terms of length, breadth, and depth (284b24), then says: Or again we may connect them with the various movements, taking principle lo mean lhal part, in a thing capable of movement, from which movem?nt first begins. Growth slarls from above, locomotion from the right, sense-movement from m fronL (284b25-29)

The conclusion toward which Aristotle is aiming is that since "the heaven is animate and possesses a principle of movement, clearly the heaven must also exhibit the above and below, right and left" (285a29). . the (12) JA 4, 705a28-b9, tr. Preus, 1981. Cf. the Gael. statem� nt (quoted in previous note) that "gro\vth starts from above." Other passages which follow the same line of thought include PA 11.10, 656a1 1-13; and Juu 1 , 468a10, which makes even more of the contrast between man and plants. Cf. PA 1 1 1.IO, 672b22-24, for "up as origin." For the analogy of mouth and root, see also de An. I I . I , 412b3; 4, 416a4; ��on� 6, 467b�; PA IV.7, 683b20; 10, 686b35; GA I l .6, 741 b33-37. See also W. K . Kraak, Aristote est-ii toujours reste fidele a sa conception que la plante se lient la tete en bas?" J\1nemosyne 3.X (1942) 251-262.

MAN AND COSMOS IN ARISTOTLE

475

human beings are paradigmatic animals, and that all other animals fall short of the perfection of human nature to a greater or lesser extent.13 But that is not exactly what Aristotle says; he does not say that man's way of being in the world determines human perceptions of the slructure of the universe. Rather, he implies that the orientation of the coordinates (I am tempted to say " Cartesian" coordinates) is independently discoverable, and that man is discoverably more in line with the universal coordinates than is any other animal. Michael's commentary on this passage makes even more obvious, without resolving, the conceptual problem it contains: he suggests that "up" is defined functionally, in terms of food intake, but if that is so, then how can we say that the "up" in the universe is really up at all, since Aristotle's universe as a whole does not take in food or excrete residues -" What, then, does "up" have to do with food? Is it that the residue of food is primarily the earthy part, that we extract from our food primarily the fire, secondarily the air, and leave most of the heavier elements? Thus the residues proceed "naturally" toward their natural places, downward. On this showing, plants would be naturally upside down, so to speak, because they are composed of a higher proportion of earth, and their "residue" (Aristotle means the fruits!) consequently concentrate the lighter elements. And we gain what we need by eating the " residues" of plants. Thus I think that the relationship between "up" and the function of nutrition is one of "accidental necessity" in Aristotle's scheme of things: the process of digestion (especially in man) is one of extracting primarily the lighter, warmer elements of food, leaving the colder, heavier elements. But that means that the natural direction of movement of the food will be progressively more and more "down", so the intake is naturally up, excretion naturally down. c) Another explanation of man's erect stature turns directly on the relationship between up and heat. This sort of argument appears particularly in the Paris of Animals: Of all animals, man has the largest brain in proportion to his size; and it is larger in men than in women. This is because the region of the heart and of the lung is hotter and richer in blood. This again explains why man, alone of animals, stands erect. For the heat, overcoming any opposite inclination, makes growth take its own line of djrection, which is from the centre of the body upwards.15

(13) This is apparently how Simon By!, for example, takes this sort of passage, Bruxelles, 1975, p. 305. (14) We recall that according to Aristotle the Pythagorean universe consumes vo'id (Physics IV.6, 213b23 ff). Plato's universe eats its own excrement (Timaeus 33c). (15) PA 11.7, .653a27-32: Ogle/Barnes. The relationship between heat and man's erect posture is mentioned in passing by Roussel, REG 1980, p. 45. PA 11.7 is also where Recherches sur les grands lraites biologiques d' Arislole : sources ecriles el prejuges,

477

A. PREUS

MAN AND COSMOS IN ARISTOTLE

The erect stature of man is again explained in terms of the natural places of the elemental powers-heat "belon gs" u p. cold down. A�othe � . . passage, in PA 1 1 1.6, confirms the relat10nsh1p between the , food . explanation and the "heat" explanation of man's erect stature: " Heat promotes growth, and abundance of blood is an indication of heat. Heat, again, tends to make the body erect; and thus it is that man is the most erect of animals, and the vivipara more erect than other quadrupeds. " 16 One of my students called this the "air head" theory of man's erect stature; if you try to imagine the mechanics of Aristotle's expla nation, it . does seem (at least at first glance) to require that human bemgs have heads full of hot air, quadrupeds less so, and snakes least. How else could heat lift up the human race? One might say that the student's response results from the imposition of our mechanics on Aristotle's science· that's true but does not solve the problem. How does Aristotle's system p �rmit heat to lift up this all too solid flesh without generating containers of hot gasses, hot air balloons? I submit that the differences in "heat" between man and the other animals are not, from Aristotle's point of view, simply matters of the quantity of heat, either in terms of temperature or in terms of . kilocalories. We need to remember that the very notwn of lemperalure in the sense of a measured scale is a modern idea; Aristotle did not have and did not envisage a thermometer. Furthermore, the idea of measuring a quantify of heat, in the sense of a calorie or in the sense of a British Thermal Unit, is intensely dependent upon prwr measurements: those of the temperature and the volume of specific materials, and their capacity to change the temperature of other materials. . We need to recognize that when Aristotle says that the. " regwn of the heart and lung is hotter" in human beings, that word "hotter" may be a qualilalive rather than a quanlilalive judgement. I rather 1magme that his theory is not refuted by simply pointing out that birds run a higher normal temperature than human beings." But if we understa nd . some of these seemingly quantitative j udgements as rather ,qualitative in character, then they become both consistent with and explanatory of

the rather obscure explicity qualitative passages; I have in mind particularly GA 1 1 .3, 736b30 ff., discussed below.

476

�

.

Aristotle says that males have more sutures in their skull than females, because their brai�l is larger, demanding more ventilation. If the brain is too moist, it will freeze the blood, if it is too dry, it will fail to cool it, which causes "disease, madness, and death." For other explanations of madness in Aristotle, see my "Aristotle on Healthy and Sick Souls," The Monist 69.3 (July, 1986) 416-433. (16) PA l l l .6, 669b5-6, Ogle/Sames. (17) Aristotle might accept the idea that birds have a higher temperature as an explanation of why they can fly; he might more readily accept a hot-air theory than a higher metabolism rate theory of bird flight.

d) A rather different mode of explanation of man's erect stature relates it to human thought (a notion notably absent in the passages cited above), and consequently to man's "divine nature. " This idea appears especially in Paris of A nimals IV: For of all animals man alone stands erect, in accordance with his god­ like nature and substance. For it is the function of the god-like to think and to be wise; and no easy task were this under the burden of a heavy body, pressing down from above and obstructing by its weight the motions of the intellect and of the general sense. 18

A train of thought of this kind is, I suppose, comprehensible but essentially alien for us. We can understand it, in a way, because we are familiar with an ascetic tradition in which fasting and slenderness are associated with mystical intellection, but we find this sort of argument alien because it puts the association between lightness and intellection into a biological rather than a religious context.19 In order to interpret the theory of mind in this sort of passage we would have to turn not only to Metaphysics A but also to De Anima I I I and Nicomachean Ethics X . For now, an examination of the cosmic importance of mind would lead us too far astray; we can, however, notice that there is a clear enough relationship between this explanation and the previous one, the one I called the "air head" theory. That is to say, this passage clearly enough shows that the human capacity of mind is related significantly to the quality of the heat in human beings; lightness is associated on the one hand with high-quality heat, and on the other with high-quality intellection. We may say that from a teleological point of view, human thought requires a physical base which has plenty of tempered high­ quality heat; the presence of that sort of heat causes the body to be lifted up and to walk around in an erect position. Consequently the body is aligned with the vertical axis of the universe, the best relationship with the universe which a body can manage and still have its feet resting on the ground.

(18) PA IV.10, 686b26-28, tr. Ogle/Barnes. 689bl 1 : consequently all other animals are "dwarflike". 690a28: consequently also man has the largest foot in proportion to the size of the body. Byl (op. cii., p. 305) remarks that even though Aristotle was unacquainted with the kangaroo, he did know the rabbit and hare. We may reply that one may easily take the "paws" (phalanges) of rabbits and hares as their feet, and those are not large. If we trace out homologous bone to homologous bone, then the "heel" (tarsus) of the horse is quite a way up the back leg; the horse runs essentially on one toe on each foot. Antelopes have proportionately the longest metatarsus bones of any animal, including the kang-aroo. See Preus (1981) p. 40. (I 9) Although we can all think of more or less ascetic philosophers - Kant comes to mind we are also familiar with some who were heavy enough to disprove what appears to be A;istotle's gener�lization in this passage - Hume would be a good example.

479

A. PREUS

MAN AND COSMOS IN ARISTOTLE

e) An explanation of man's erect posture which Aristotle does not offer is the one which Galen gives, that it frees the hands for their multiple functions for the human being.20 This sort of explanation would, I think, smack of Anaxagoras' notion, refuted in PA IV.10, that man is the most intelligent of animals because he has hands; rather, man has hands because he is the most intelligent animal. In fact Galen comes close to falling into the Anaxagorean error himself.21 Nor does Aristotle recognize Xenophon's explanation, that we can . thus see objects above us more easily, and thus avoid danger.22 No, man has hands because he is most intelligent, and being most intelligent causes an erect stature; nature makes use of the erect stature to provide this tool of tools.23 Assuredly this is not biological functionalism; rather, this way of thinking belongs to another mode of thought entirely, more in tune with the cosmic teleology of Metaphysics A than with anything we readily recognize as a scientific appreciation of the world. Before I get in too deep, let us look at two other relationships between biological and cosmic reality, the analogy of pneuma and aither and the relationship between cosmic and biological periods.

naturally moves with a circular motion, a body distinct from fire (pace Anaxagoras); that body will be ungenerated, undestroyed, eternal and . divine. Everyone agrees that the immortal is divine, and the heavens are observed not to change at all.

478

2) PNEUMA AND AITHER Explicit reliance on the notion of aither is relatively rare in the Aristotelian corpus. Neither Metaphysics A nor GA 1 1 .3 uses the word aither; rather the one talks of "the body (crwµ�) which moves in a circle" (I073a31) and the other, mentioning a crwµ�, refers to To TWv iXcrTpwv crTotxdov. Even in the De Caelo Aristotle usually uses a descriptive phrase rather than the word aither.24 Still, there is some justific�tion for using word ailher in this context. In the beginning of the de Caelo, Aristotle argues that to explain the motion of the universe we must posit that there be some "body" which ·

(20) Galen, de Usu Partium 1 1 1.3, May 160, for example. (21) Idem l ll . l , May 154. (22) Xenophon Mem I.4. 1 1 , reference thanks to By!, p. 128. (23) PA IV.JO, 687a6ff. (24) De Caelo 1.2, 269a31 : ·ni:; oUata: aWµ
·n 1tixpCt. -riX aWµa.-ra; -r&. 8Upo xlXl ne:pl �µiii:; E-re:pov xe:x.wptaµtl:vov, -roao6-rci> ·nµLw-ri:pct:v lx_ov -ri)v q:iUatv 5at::i1te:p &ipScrT'Y)X£ -rWv Sv't'a:U0a: nAe:tov. I .3, 270b3: TO 7tp&-rov -rWv crwµ&.-roov {not 7t€µ7tTov GOOµa:). For a discussion of the development of Aristotle's theory of this body, see David 1-Iahm, "The Fifth Element in Aristotle's de Philosophia: A Critical Reexamination", in Anton and Preus, eds., Essays in Greek Philosophy 11, Albany 1983, pp. 404-428. See also J. Thorp in Phoenix XXXVI (1982) 104-123 for an interesting discussion of how ailher

lights up, according to Aristotle.

The name, too, of that body seems to have been handed down right to our own day from our distant ancestors who conceived of it in the fashion which we have been expressing. The same ideas, one must believe, recur in men's minds not once or twice but again and again. And so, implying that the primary body is something else beyond earth, fire, air, and water, they gave the highest place the name of aither, derived from the fact that it "runs always" for an eternity of time.2s

The parallel between this passage and the first of the two I quoted at the beginning from Metaphysics A is obvious; only the name "aither" has been omitted from the Metaphysics passage. Aristotle has persua­ ded himself that the idea of the divinity of the heavens is one which persistently recurs throughout human history, and that it is one of the few leading ideas which regularly survives the periodic catastrophes to which human history is subject. Occasionally Aristotle uses the word ailher as a convenient shorthand for the idea that there is some crwµoc or ousia which is superior to all the terrestrial elements, moves in a circle, and is identified with the heavenly spheres. A great deal can (and has26) been said about the role of pneuma in Aristotle's explanation of life and soul in the biological works. This is not the place to review all of that; I only want to say a few things about Generation of Animals I 1.3, 736b30 ff: ... It remains that mind alone come in additionally from outside and alone be divine; for bodily activity has no share in its activity. Now the poWer of every soul seems to share in a body different from and more divine than the so-called elements; but as souls differ in value and lack of value, so too this sort of nature differs. All have in their semen that which causes it to be productive; I mean what is called "heat". This is not fire nor any such power, but it is the pneuma wrapped up in the semen and in the foam­ like, more precisely* the nature in the pneuma, which is analogous to the element of the stars. Hence, whereas fire generates no animal and we do not find any living thing forming in either solids or liquids under the influence of fire, the heat of the sun and that of animals does generate them. Not only is this true of the heat that works through the semen, but whatever other residue of the animal nature there may be, this also has still a vital principle in it. From such considerations it is clear that the heat in animals neither is fire nor derives its origin from fire.21

(25) De Caelo 1.3, 270b16-23, tr. Stocks/Barnes, with minor revision. Cf., e.g., Gael. 11.7, which says that the stars are composed of that element. (26) See especially M . Nussbaum, Aristotle's De Mofu Animalium, Princeton 1987, pp. 143-164, with her citations; G. Verbeke, "Doctrine du pneuma et entelechisme chez Aristote'', in G. E R. Lloyd and G. E. L. Owen, Aristotle on Mind and the Senses, Cambridge 1978, pp. 191-214. (27) CA I 1.3, 736b30-737a6, my translation, partly relying on Platt/Barnes, partly on Balme. See b. M. Balme, Aristotle's De Pariibus Animalium I and De Generalione

480

A. PREUS

MAN AND COSMOS IN ARISTOTLE

One quick comment: this discussion of pneuma occurs immediately after a striking statement of the separate origin of mind. Although the explanation in the passage is not developed simply in order to account for the coming into being of mind in the human soul, it is likely to be relevant for such an account.28 In this passage it is assumed that there are (at least) two kinds of heat, the heat of fire and vital heat; we may conclude from the passage that (vital) heat causes animal generation,29 and that (vital) heat pccurs notably in pneuma. (The "body" ("wµoc-ro,) at 736b30 has to be pneuma.) Because they contain (animal) pneuma, various animal residues also contain vital heat. Aristotle has in mind, for example, the feces within which many insects and so on are supposedly generated (in so-called spontaneous generation).30 This pneuma contains some "nature" (the heat?) which is "analo­ gous to the element of the stars," i.e. aither. How should we understand the word ocv
might be present in the psychic pneuma. Peck remarks32 that de Anima I I .7, 418b6-10 states that the transparency of air, water, and various solids is due to the presence in them of the "same nature as also in the eternal bodies above. Light is the activity of this transparent qua transparent." One might imagine that Aristotle thinks that aither is present in water in that water is transparent, and that pneuma is present in water in that it is warmed by the sun or other heavenly body. What is the relationship between the light of the sun and the heat of the sun? That may be one way of characterizing the relationship between the aither in the sea and the pneuma in the sea.33 In Generation of Animals I I I . ! ! we get a somewhat different account of "spontaneous" generation:

The pervasiveness of pneuma and consequently of vital heat (and, in a way, soul) tends in a direction rather away from the hylomorphism of de Anima: there, Aristotle criticized the idea that "the soul pervades the whole universe, whence perhaps came Thales' view that everything is full of gods," on the ground that describing " fire or air as living creatures is highly irrational , and yet to refuse to call them living creatures, if there is a soul in them, is absurd. "35 Indeed, it seems to me that the argument in de An. I.5 is inconsistent with both GA I l.3 and II I . 1 1 , since in both GA passages Aristotle is essentially asserting that soul is present in hot air (pneuma). To be sure, in GA I I .3, generative power is limited to animal heat, which might be interpreted as a (separated) organ of the animal, a capacity of generating which belongs ultimately to the organic unity of the generating animal. But . GA I l l . 1 1 cannot be interpreted in that way, since the pneuma in sea water is not at all said to have an animal origin. There may be two sorts of heat at work in I l l . I I : the vital heat present in all air (762a20) is

I (with passages from 11.1-3), Oxford, 1972, especially his note, pp. 1581 65. At*, the translation follows Paul Moraux, "A propos du N00:E 0YPA0EN chez Aristote," Auiour d'Arisiole, Louvain 1955, p. 255; see Balme p. 163. Roussel discusses this passage {briefly} in REG 1 980, pp. 68-69. (28) Balme's note to the passage gives a plausible enough interpretation; I will not duplicate it here. (29) See Balme on this, as well as Nussbaum. (30) HA V.19, 551a4 ff: " ...others grow in decaying mud or dung; others in timber, green or dry; some in the hair of animals; some in the flesh of animals; some in excrements: and some from excrement after it has been voided, and some from excrement yet within the living animal, like the helminthes or intestinal worms." (31) A. Preus, Science and Philosophy in Arisloile's Biological Works, I·Iildesheim 1975, pp. 86-89. Nussbaum MA p. 161-162, arguing that an identification of pneuma with ailher will not solve the problem of purposive movement in MA 10, claims to be in

agreement with Balme, Moraux, and Riische in rejecting "any facile identification of pneuma with ailher" on the basis of our passage. I would now say that I am not attempting to "identify" pneuma with ailher, but rather to show that they are closely related in Aristotle's mind in this passage, and perhaps elsewhere. (32) A. L. Peck, Aristotle: Generation of Animals (Loeb), note to 7�7al, and see also his appendices. (33) Thorp's comments (Phoenix 1982) on celestial light and heat are helpful here. (34) GA 111.1 1 , 762al9-22, my translation. The last few words are of course an unattributed reference to one of Thales' reported opinions, directly ascribed to Thales at de Anima 1.5, 4 1 1 a7. For a good general discussion see James Lennox, "Teleology, Chance, and Aristotle ' s Theory of Spontaneous Generation," Journal of lhe llislory of Philosophy xx (1982) 219-238. (35) De An;mp 1.5, 41. la7 ff.

481

Animals and plants are generated in earth and in water because in earth water is present, in water pneuma is present, and in all pneuma psychic heat is present, so that in a way aH things are full of soul .34

Animalium

482

MAN AND COSMOS I N AR!STOTLE

A. PREUS

enclosed in a frothy bubble, and the "heat of the warm season in the environment" (762b l4) which "combines by concoction out of sea-water or earth. "36 But however the passage is interpreted, Aristotle has taken a direction that gives him an explanation of the generation of sea animals at the cost of part of his hylomorphism; one may say that he gains in cosmic integration what he loses in animal integrity. Returning to the passage in I 1 .3, we note in addition that the pneuma differs in value (,tµt6'�'') in respect to the particular psychic heat which may be present in it. Here the differences between varwus lots of vital heat are explicitly qualitative rather than quantitative; we do not have to claim that the human semen is actually holier than any other semen, only that its heat is qualitatively better. Reading a passage like this with a modern scientific agenda, we are left with an uneasy feeling, because we cannot envisage how one could go about verifying the hypothesis empirically. Still, we do get an impression of the sort of relationship which might exist between the various sorts of "heat" in the universe; firey heat, and aitherial heat might be so different qualitatively, in Aristotle's mind, that they don't fit on the same scale, and are thus analogous rather than continuous. At the same time, we might want to think that the "lower" kinds of heat are ultimately derived from the "higher." Aristotle 's denial of the thesis that the heat in the pneuma is derived from fire (i.e . , his denial that a higher form of heat might be derived from a lower) may be part of a larger project of refuting the theory of the primacy of fire in the cosmos and in life. The name of Anaxagoras is often associated with the theory that the celestial bodies are fire, in the "cosmic" books;37 Democritus is credited with identifying the soul as fire in the de Anima.38 But Aristotle is more consistently opposed to the thesis that the stars are fire than he is to the thesis that vital heat is a form of fire. In PA 1 1 .2, when he provides what looks meant to be a thorough analysis of the various senses of the word "hot," he does not at all distinguish between vital heat and the heat of fire. In fact, he seems to assert that the process of digestion is an in'stance of combustion (eµ7te7tupeucr80.t, 649a27). Later in the book, at PA 11.7, Aristotle again makes a much less rigid distinction between psychic heat

(36) Allan Gotthelf suggested at the conference that the water-pneuma is enclosed, and the sun-heat makes the external envelope, according to I I 1 . 1 1 . That certainly is a possible interpretation of the passage, though I would want to suggest that the heat of the pneuma in sea-water must also be derived from the environment and the sun, since it is explicitly said that it is not derived from animals. (37) Gael 1.3, 270b24; !11.3, 302b4; Mete 1 .2, 339b22; 11.9, 36%14 (in explaining lightning}. (38) De An. 1.2, 404a I .

and fire than that made in GA , for example. of the brain, he notes that:

483

Dealing with the function

Some writers assert that the soul of an animal is fire or some such power. This, however, is but a crude assertion; and it would perhaps be _ some such body. better to say that the soul is incorporate in

Noting the pervasive importance of heart for the functioning of the animal constitution, Aristotle goes on: To say then that the soul is fire is much the same thing as to confound the auger or the saw with the carpenter or his craft.39

Although Aristotle does not here argue that the "body' '. of the soul is not fire, he is seriously at pains to argue that the soul itself is not . identical with fire. And he does not assert that fire LS the body of the soul, only that "perhaps it is better to say that the soul subsists in "t"OLOU"t"W "t"LVL crwµ.ix-rL. " I� Juv. and Resp. Aristotle makes a clearer distinction, talking for example at Juv. 4, 469bs ff, about "some natural connate heat" which . . distinguishes between life and death; the heart is the hearth for this connate heat, and nutrition and the air may maintain the " � atural heat" (6, 470a23). Resp. 8 tells essentially the same story; lll fact respiration is taken to be primarily a cooling mechanism, �nd th � Eµ.cpu-rov _ non-breathmg animals (9, meuµe< is the analogous cooling mechanism m 475a9). . Of course the distinction between the heat of fire and vital heat may be related directly to Aristotle's general theory that various lots of heat differ qualitatively, in terms of their value, quite as much or even more than quantitatively, in any measurable way. Aristotle doubtless thmks of the heat of fire as "low-quality" heat, which is why it cannot generate anything living. The heat of the sun and of the other celestial spheres is, obviously, very high quality heat, and consequently does generate, . either by itself (so-called "spontaneous" generation), or m concert with the parents ("man and the sun generate the offspring").'° Returning to the question of "mind": Aristotle has some rather obscure notions of relationships between pneuma, aither, heat, and light, on the one hand, and mind on the other. To be sure, vou<; 8upe<8ev and VOU<; xwptcr,6<;, but the theory that mind is absolutely immaterial is a little crinkly around the edges. Pneuma, in the GA and MA at least, has the ,

aWµix.

\

,

(39) PA 11.7, 652b7-13,

tr. Platt/Barnes, with "power" for 8Uvixµti;; and "body" for

(40) Metaphysics A. 5, 1071a 12: "The cause of ma? is the elements in ma�· ·· and the . external cause ' whatever it is, e.g. the father, and besides these the sun and its oblique course, which are ... moving causes." Cf. GC 11.10 on the "oblique .course" as moving cause. At GA I . 16 , 716a16, Aristotle remarks that the mythologists call the earth 'mother' and the sun and heavens 'father'.

484

A. PREUS

function of carrying information from one location to another -" We are tempted to relate this function of pneuma to the idea of logos, although that is not how Aristotle phrases it. Rather, pneuma and the psychic heat are said repeatedly to be an ?t.px.�- of the animal, of generation, of movement. Peck makes the case that Aristotle's pneuma functions not only in generation and movement, but in perception as well, although this part of Aristotle's theory is far from worked out in any detail. If pneuma carries information within the body, for sensation and movement, it functions somewhat as we would say the nervous system functions. Indeed, once the nerves were discovered, they were believed (even by Galen) to be hollow tubes full of pneuma, so persuasive was the Aristotelian explanatory system. But if pneuma performs many of the functions of the nervous system, we would suppose that it must have a close relationship with mind. And, as we noted above, man stands upright because he has more or better vital heat than other animals, and he needs to have that sort of vital heat because he is intelligent. Finally, is there more than the surface significance in that famous passage, de Anima 1 1 1 .5, 430a l 4 ff? This sort of mind (passive) is such because it becomes all things, but the mind which is what is because it makes all things (active) is a sort of positive state (��i.:;) like light; for in a way light too makes potential into actual colors.

Is the relationship between active thought and light more than metaphorical? I only want to call -attention to the connection which Aristotle makes here; I do not have a developed interpretation. 3) BIOLOGICAL AND COSMIC P ERIODS We come now to a passage which is fundamental for understanding the relationship between Metaphysics A and the Generation of Animals, the passage at the end of GA IV. Aristotle is explaining why the period of gestation in varies in different animals as it does. He starts by saying that the period of gestation bears some relation to the lengh of life, and then corrects himself and says that the larger the animal, the longer its gestation period is likely to be. But that (reasonable enough) explanation does not satisfy him; he goes on to find cosmic significance in the periodicity of the generation of animals. We find, as we might expect, that in all animals the time of gestation and development and length of life aim naturally at being measured periods. By a period I mean, e.g., a day and night, a month, a year, and

(41) Cf.

GA

IV.3, 767b21;

MA

10, 703a5 ff.

MAN AND COSMOS IN ARISTOTLE

485

the times measured by these, and also the periods of the moon... The moon is a first principle because of its connexion with the sun and its participation in its light, being as it were a second smaller sun, and therefore she contributes to all generation and development. For heat and cold varying within certain limits make things to come into being and after this to perish, and it is the motions of the sun and moon that fix the limit both of the beginning and of the end of these processes. Just as we see the sea and whatever is fluid* settling and changing according to the movement or rest of the winds (7tve:Uµ1x:rct.), and the air and winds agains according to the course of the sun and moon, so also the things which grow out of these or are in these must follow suit. For it is reasonable that the periods of the less important should follow those of the more important. As for the revolutions of the sun and moon, they may perhaps depend on other principles. It is the aim, then of nature to count the coming into being and the end of animals by the numbers of these higher periods, but nature does not bring this to pass accurately because matter cannot be easily brought under rule and because there are many principles which hinder generation and decay from being according to nature, and often cause things to fall out contrary to nature.42

The first thing we should notice in this passage is that the direcl cause of genesis and destruction of animals and other living things is "heat and cold varying within certain limits (µ<x.r< cruµµ•�ri�,)" (777b28). In other words, the claim is made that the orbiting sun and moon measure the life-span of entities by altering the "heat". By now we know enough about Aristotle's theory of "heat" that we should not too quickly say that the sun and moon simply c.hange the "temperatu­ re", in our sense of the word. The issue is one of celestial heat, which is the higl)est quality of heat available in the universe. Of course we would agree that the. influence of the sun upon environmental temperature-day and night, summer and winter, is obvious. As Aristotle makes clear in the Generation and Corruption and elsewhere, it is the "oblique cycle" which actually brings about the major differences; i.e., it is the fact that the sun is more toward the north in the summer, and toward the south in the winter, that makes the seasons. But he is talking about the quality of the environmental heat as much as its quantity or degree. Aristotle appears to argue, without the sort of evidence we would expect, that the moon too has an influence on environmental heat. On the one hand, he has some empirical observations which indicate that

(42) GA IV.10, 777b16-778a9, tr. Platt/Barnes. At* I follow Peck's translation, which is more literal and more informative. Compare GC 11.10, 336b10 ff: The times, i.e. the lives, of the several kinds of things have a number by which they are distinguished; for there is an order for all things, and every time (i.e. every life) is measured by a period . Not all of them, however, arc measured by the same period. Not all of them, however:, arc measured by the same period, hut some by a smaller and others by a greater one; for to some of them the period which is their measure is a year, while to some it is longer and to others shorter.

I

I

486

487

A. PREUS

MAN AND COSMOS IN ARISTOTLE

there are life cycles which follow the cycles of the moon: the clearest example is probably that of the sea-urchins, whose cji6v (which Aristotle takes to be not an egg but fat) is largest at the full moon, "not because the animal eats more then, but because the nights are milder due to the light of the moon". Sea urchins also do better in summer, because of the warmth.43 He knows that the woman's menstrual cycle tends to be the same length as the lunar cycle; he imagines that most women tend to be synchronized with the cycle," because they should be synchronized with the supposedly warmer and colder parts of the month, according to the full and new moon. Because Aristotle does not know about the gravitational pull of the moon, he does not have an adequate explanation of the moon's influence on bpdies of water. He thinks that the changes of the moon bring about changes in the winds through heat and cooling, and the winds in turn bring about motions of . liquids. Actually, I wonder whether Aristotle is not including both senses of the word pneuma in this passage. It does seem to me possible to read the passage that way. Wouldn't the heating and cooling of the environment have an influence on the "rest and movement" of all pneumata, whether winds, breaths, or connate air?45 " It is reasonable that the periods of the lower follow the periods of the higher (xi:x't"OC A6yov yOCp &.xoAou6e:Lv xi:x.L T0Ci:;; -rWv &.xupo-rS::pc.uv 7tE:fn63ouc; -ri:xi:'i:;; TWv xupwT't"e:pa. or xup<6npet than others, then the lower should be expected to follow the leadership of the higher. The principle thus enunciated is, in a way, a j ustification of •the anthropocentric or anthropomorphic explanations with which the biological books are replete: it is absolutely to be expected that the lower animals be, in a way, more or less defective representations of the best

terrestrial creature, man. We should similarly expect t.hat man is but a defective representation of the entities which are above. We can of course trace the analogy back to Heraclitus,46 and we recall the trenchant criticism which Plato's Stranger makes of anthropocentric classifications, at Statesman 263d (the rational cranes might "invest themselves with a unique and proper dignity and classify the race of cranes as being distinct from all other creatures"). Probably Plato and Aristotle agree in not really having an ultimately anthropocentric mode of explanation, but rather taking as source and model of all understan­ ding of the myriad creatures the structure of the universe as a whole. If there really are "epistemological breaks" in the history of thought, we should certainly like to locate one of them in the vicinity of this passage. We generally do not have much difficulty with Aristotle's theory of the four causes so long as we talk about bronze statues, the human hand, or the elephant's nose, but as we read these passages from Metaphysics A, the de Caelo, and GA IV, we are struck by the strangeness of the application of the four causes to the universe as a whole. In the first place, we do not know what to do with the claim that "all things are ordered together to one end . " Local final causes can be understood according to biological functionalism; universal final causes are more puzzling, particularly when they are Aristotelian final causes. Our religious teachers, and Plato too, claim that the universe as a whole is an aesthetic object for God.47 But Aristotle's God appears not to have the world as an object for thought, let alone as an object for aesthetic appreciation. How does an army contain the good? Yes, as the excellence of the general and of the organization which depends (partly) on the excellence of the general, but that's an inadequate answer. The goodness of an army belongs to its functional capacity as a fighting force; its excellence is judged by something outside itself, by the environment within which it functions. And the universe has nothing but God outside itself, according to Aristotle, and no environment by which it is judged. Internally, the teleogical order is just as puzzling to us. We can make sense of the notion that the various kinds of animals exist for the sake of the whole environment within which they function; even though most biological scientists will chide us for this way of thought, we can readily imagine that bees exist partially in order to _p ollinate flowers,

(43) PA IV.5, 680a31-34. (44) HA VI I.2, 582b1 (without modifiers); GA 11.4, 738a20: "the period tends to return during the waning of the moon. This we should expect, for the bodies of animals are colder when the environment happens to become so, and the time of change from one month to another is cold because of the absence of the moon." At GA IV.2, 767a l , he adds that "as the sun makes winter and summer in the year as a whole, so does the moon in the month. " In an earlier publication I speculated that Aristotle may have made a hasty generalization about the time of month at which women menstruate, on the basis of too little information. I now think that the generalization was pre-determined by the theory into which the observation was to be fitted. (45) Earthquakes sometimes coincide with an eclipse of the moon because the wind (pneuma) turns off into the earth: Meteorology 11.8, 367b20 ff; cf. Problems XXVl.18, 942a22.

t

I

I.

(46) Plato, Hippias Majoe 289b, apparently citing Hecacleitus (DK 883), "The wisest of men appears to be but an ape in comparison with a god ... " Cf. DK B79. (47) Genesis 1.31: "And God saw everything that had made, and behold, it was very good." Timaeus 30d: "For the deity, intending to make this world like the fairest and most perfect to intelligible beings, framed one visible animal comprehending within itself all other animals of � kindred nature".

i

489

A. PREUS

MAN AND COSMOS IN ARISTOTLE

that scavengers exist partially in order to rid the environment of carrion. But what sense can be made of the idea that the entire terrestrial ecosystem exists for the sake of the lunar sphere, that everything within the lunar orbit exists for the sake of the solar sphe �e, that the solar system exists for the sake of the sphere of the fixed stars? And I remind you that the material cause of the visible entities varies qualitatively as one proceeds from the center of the earth to the outer limits of the universe. That qualitative variation is most readily evidenced by the increasing regularity of the motion of entities as one proceeds outward, the increasing unity of the originating cause, the decreasing variability which might be introduced by the imperfections of the lower forms of matter (remembering especially GA IV.10). The "epistemological break" may be a rather large one, something rather like a chasm in fact. We modems expect that the mailer of the universe is fundamentally identical; the current story is that it all reduces to the same few basic quarks. And we suppose that the physical laws governing the construction and motion of material things are also universally uniform. . Aristotle does not talk of universal laws of nature, nor does he think the matter uniform. Although both we and Aristotle suppose the physical universe to be one and unified , for us that unification depends upon the laws governing the behavior of the ultimate matter (in that, we follow Democritus far more closely than Aristotle); Aristotle supposes the unification of the universe to be essentially formal and final. In order to make his formal and final unity work, Aristotle needs to have some permeation of qualitatively superior matter into the sublunar world; thus the pneuma and the vital heat in the pneuma which makes possible the functions of life are said to be related, in some rather obscure way, to the celestial aither. To the modern mind, understan­ ding the presence of pneuma in the terrestrial environment as a penetration of the celestial and the divine into the sublunar realm looks like a touch of mysticism, which is to say it is mysterious to us.

problematic details, let us look briefly at the heart of the matter, at 6, 700b30:

488

A Brief Comment On MA and Met. A. The de Motu Animalium is a short treatise both very informative about many points in Aristotle's theory of animal local movement and terribly puzzling in the analogy drawn between the movement of animals and the movement of the heavens.48 Putting aside the many

{48) Martha Nussbaum writes (MA p. 135): "We imagined that we knew what Aristotle was up to, and that his preoccupations were ours. And now the meticulous analyst is found constructing a picture of the heavenly bodies on the basis not of careful observation and measurement, but of popular superstition." She goes on to describe Aristotle as appearing here to be the '"religious man' who fights for the right of supernatural to remain unexplained, of the marvelous to remain unchallenged by

It is clear that the eternally moved by the eternal mover, and the individual animal, move in the same way in one respect, but in another respect differently, in that the former move eternally, but the movement of animals has a limit.

Aristotle goes on to say that the eternal mover is eternally valuable, and implies that is why things which are moved directly by it move continuously, without stopping; animals, on the other hand, move in virtue of orexis, which is in a sense a moved mover, and has a limit. Animal movement (chapter 7) may be analyzed in terms of practical syllogisms, and those syllogisms (and the situations they represent) have "limits", namely the specific goals which are achievable by action. Thus Aristotle says that the most significant difference between celestial and animal movement is in terms of the final cause of their movement: the celestial bodies move directly for the sake of the absolutely best, animals move for the sake of the relatively good. The theory may well be elucidated by quoting the continuation of a passage from Metaphysics A cited at the beginning: It is as in a house, where the freemen are least at liberty to act randomly, but all things or most things are already ordained for them, while the slaves and beasts do little for the common good, and for the most part act randomly; for this is the sort of principle which constitutes the nature of each.49

Aristotle tell us that free persons do not act randomly, because they are determined by their goals, not just their personal goals, both those of the entire community; slaves have only a limited understanding of what ought to be done, and consequently are relatively little determined by their concept of the good. The passage may well be put in conjunction with this one from the MA : The animal must be understood as constructed like a well-governed city, for once order is established in the city, there is no need for a separate monarch which has to preside over each activity; each individual does his own ordered task, and one activity follows another through custom. In animals this happens by nature . . .00

reason .." I differ with Nussbaum on two points here: a) we have no reason to expect Aristotle ever to measure anything (he never or almost never even suggests measuring something}; b) Aristotle's picture of the heavens is not really based on popular religion (even less superstition), despite his claim to have derived it thus. Aristotle is more nearly engaged in the establishment of a religious dogma than in the retailing of an existing faith. (49) Mtil. A. 10, 1075a19. G. E. R. Lloyd pointed out in the conference that the Ross/Barnes translation misleadingly says that "the freemen are least at liberty to act as they will" and that "t1he slaves... for the most part live at random" when the operative phrase in both places is a ..� e..uxev. (50) MA 10, 703a29-34, my translation.

i

' ' ' !

! •

490

B1ouxm::. LOGIOUE Er M2TAPHYSIQU£

A. PREUS

The macro-level of the cosmic army or household of Metaphysics A is at least partly reflected on the micro-level in the animal as city, for we can ask the same question asked in the Metaphysics: how is the good contained in the animal, as the goodness of its leader or the arrangement of its parts? Probably in both ways ... for the soul is that for the sake of which all the parts of the animal function, and the excellence of the soul is reflected in the excellent functioning of the parts, "once order has been established". Last Word. Are we in a position to understand Aristotle's cosmic teleology, or not? Our religious tradition tells us that the world is good because God made it that way; cosmic teleology exists in our religious, if not our scientific, imagination through the notion of design transferred from the human realm to the cosmos. But Aristotle's universe is eternal, not created, and therefore not designed. Nor is it good simply because it is beautiful, although it is beautiful. The answer, such as it is, is to be found at the beginning of Book I I of the Generation of A nimals: Now some existing things are eternal and divine whilst others admit of both existence and non-existence. But that which is noble and divine is always, in virtue of its own nature, the cause of the better in such things as admit of being better or worse, and what is not eternal does admit of existence and non-existence, and can partake in the better and worse. And soul is better than body, and the living, having soul, is thereby better than the lifeless which has none, and being is better than not being, living than not living.51

Being is better than not-being. There, in a word, is the heart of Aristotle's teleology. The universe is as it is because it thus maximizes existence, soul, and mind. This is a faith, if you like a religious faith, but it is not simply an acceptance of popular superstition or an unquestioned assimilation of the religious tradition. It is the cornersto­ ne of the possibility of thinking about the world in an Aristotelian way. Is it a way of thought which is available to us today,, or have we become separated from that way of thought by an unbridgeable epistemological chasm? Do we want to say that we are no longer sure whether existence or non-existence is better? Or if we do accept the premise that existence is better than non-existence, must we also be committed to a picture of the universe as thoughtfully going around in perfect circles forever? I'd like to say that we are not separated from the premise that existence is better than non-existence, and that we need not be committed to a closed universe to accept that premise; in a way, our universe is all the more excellent than Aristotle's for being more vast and no less regular. But that is quite another story altogether. (51) GA

II.I, 73l b24-30, Platt/Barnes.

clm

Am=

8eminaire CNRS-N.S.F.,

1987

Editions du CNRS, Paris, 1990.

HOMEOMERES ET ANHOMEOMERES EN THEORIE BIOLOGIQUE D'ARISTOTE A AUJOURD'HUI RENE THOM

S'il est une partie de l 'ceuvre d'Aristote qui, actuellement, semble irremediablement perimee, c'est bien sa theorie de l'organisation biologique; elle repose - on le sait - sur le couple des notions : homeomere-anhomeomere. Une telle distinction, fondee sur l'examen phenomenologique des organes mis en evidence par dissection - sans autre moyen que l'examen visuel - pourra sembler, a nous modernes, tout a fait rudimentaire. Pourvus du microscope, nous savons depuis plus d'im siecle que l'etre vivant se decompose en cellules, chaque cellule constituant en quelque sorte l'unite indivisible, l'atome du vivant. Mais ii n'est pas sur que cette connaissance frappe d'invalidite toute tentative d'expliquer theoriquement - de formaliser - la structure visible de l'organisme pluricellulaire. Nous pensons au contraire que les structures grossieres mises en evidence par l'examen visuel constituent un "niveau hierarchique d'organisation" - notion obscure qu'il nous faut certes elucider -; on montrera qu'un tel niveau est pourvu d'une certaine autonomie - laquelle permet une description intelligible independante des structures fines cellulaires ou subcellulaires. Cette tache s'inscrit dans le droit fil des considerations du De Partibus Animalium (note P.A.) et du De Generatione Animalium (note G.A.), elles-memes issues des considerations plus generales de la Physique () et de la Metaphysique (Met.). I. Les Homeomeres. Hypothese sur l'origine.

La definition et les proprietes generales des homeomeres et anhomeomeres sop t donnees au livre II de P.A., apres le livre I qui, lui,

-...

492

RENE THOM

est consacre a une critique methodologique de la dichotomie platonicien­ ne. Apparemment, aucun lien evident ne relie le cont_enu des deux liures. Nous allons ici proposer une ceriaine reconstruction de ce « chainon manquant »1. Pour cela ii nous faut revenir a la decouverte par Aristote du substrat materiel. Seton la these defendue par D. W . Graham dans son livre «Aristotle's two systems»2, }'auteur des Categories serait initialement parti de la conception d'un « atomisme substantialiste » (celle de l'Organon), oil deux predications : X est A, X est. B sont contradictoires si A et B sont deux qualites contraires d'un meme genre, X etant une entite (premiere). Pour rendre compte du devenir et du mouvement (et refuter Parmenide), Aristote aurait alors introduit le substrat (u7toxdµevov) - a savoir la matiere -, qui seul permet la consecution temporelle de deux predications d'une substance (ousia). Je proposerai ici qu'une autre motivation - plus immediate - a pu jouer dans !'invention du substrat. Aristote etait trop conscient des faits de langage pour ne pas s'apercevoir qu'une assertion telle que (NB) Ge chat est noir et blanc est parfaitement acceptable d'un point de vue semantique (et linguisti­ que) et cependant dans un univers d'atomisme substantialiste, une telle phrase serait intrinsequement contradictoire (comme elle le serait en logique Booleenne moderne - un fait dont nos theoriciens des Mathematiques Modernes n'ont pris conscience que bien tardivement)3. G'est dire que le role du substrat est non seulement de permettre le changement, le devenir, mais aussL la coexistence simullanee dans une

(1) II n'entre pas dans mon propos ici de discuter le lien relatif (ou !'absence de lien} entre le livre 1 de P.A. et les livres II-IV qui suivent. On sait les controverses entre erudits auxquelles ce probleme a donne lieu. Le livre 1 fut-ii un traite methodologique autonome rattache ulterieurement aux anciens livres ? Leur est-ii anterieur ou posterieur? · S'il m'etait permis d 'exprimer ici !'opinion d'un incompetent, je dirais qu'a beaucoup d'egards, !'introduction des homeomeres et anhomeomeres en Biologie apparait comme une tentative de sauver la methode diairetique des difficultes qui dans le domaine de ces etres concrets que sont Jes organismes animaux, ruinent l'efficacit.e de la belle methode logique de l'Organon. Par la vision generale du rOle de la nature, de la matiere, de la forme et des causes finales, le livre 1 de P.A . me semble tres proche des livres II et I I I de la Physique ; ceci conforterait !'impression generale d'un interet relativement precoce d'Aristote pour la Biologie - sans doute non anterieur aux grands traites philosophiques de la Physique et de la Metaphysique, mais de fort peu ·posterieur. (Voir les observations de P. Pellegrin sur la theorie opposee de d'Arey Thompson, note de bas de page 51 de CAA). (2) lei encore, mon incompetence ne me permet pas de juger de la validite de la these de D. W. Graham. On peut croire a une evolution « 8 la Wittgenstein» (si !'on permet cette comparaison sans doute fort inadequate). Parti d'un logicisme rigoureux, preoccupe d'abord de combattre les Sophistes, le Stagirite aura mUri en un philosophe de la nature, dont le but essentiel etait de conferer au monde sensible l'intelligibilite. (3) Voir a ce sujet mon article : « Les Mathilmatiques Modernes, une erreur pedagogique et philosophique » ? in «i' Age de la Science», juillet-sept. 1970, vol. III, Dunod ed., Paris, particulierement l'Appendice p. 235-42.

HOMEOMERES E'T ANHOM:EOMERES D'ARISTOTE

A

AUJOURD'HUI

493

entite de « parties» ayant des qualites differentes, voire contra ires. Ainsi le pelage du chat, sujet de la phrase N . B . , se divisera en une partie blanche, et une partie noire. La notion d 'homeomere permet ainsi de recouvrer un certain atomisme substantialiste clans le cas d 'une entite qualitativement composite4• lei, bien entendu, on pourra m'objecter qu'entre en jeu le caractere etendu, spatial, du substrat, et qu'Aristote a refuse avec l'obstination la plus complete, toute consideration de l'espace, pour lequel il n'a aucun mot. (On pourrait penser que 6ic11 �omx� represente l'etendue, mais il s'agit d'un hapax dans le corpus). Ge serait oublier que pour lui la plupart des substrats (sinon tous) sont des continus5. On voit done que si (B) est le corps d'un animal, en associant a chaque point x de B la totalite des « qualites» susceptibles d'etre prediquees d'un voisinage de x dans B, on definit une partition de (B) en classes d'equivalence ; une telle classe d'equivalence constituera un « homeomere ». Les exemples donnes par Aristote d'homeomeres (le sang, la chair, la moelle des os, etc.) sont des « volumes» de l'espace (des etendues tri-dimensionnelles) ; deux tels homeomeres contigus sont separes par une « cloison » qui est une surface S. On peut alors definir pour (S) une partition en homeomeres de la meme maniere, en exigeant que les voisinages de 2 points de S aient des voisinages dans B qualitativement « homeomorphes » . . . et ainsi de suite : (S) aura pour bord des courbes qui sont des homeomeres de dimension 1 , lesquels auront pour bords des points (dimension zero). Ainsi cette notion d'homeomeres conduit a munir le volume (B) d'une partition en <( varietes connexes » X i (Les slrales) qui sont des homeomeres de dimension 3, 2, 1 , 0. On retrouve ainsi l'objet connu en mathematique sous !'appellation d'ensemble slralifie. U n polyedre est un ensemble stratifie : par exemple le cube classique, 13, se decompose en : 1 strate de dimension 3 : l 'interieur 6 strates de dimension 2 : les faces 12 strates de dimension 1 : les aretes 8 strates de dimension 0 : Jes sommets (observer que 8 + 6 - (12 + 1) 1 , relation d ' Euler). =

(4) V. Cherniss (Arislolle's criticism of Plato and the Academy, p. 175, note de bas de page) a invoque le passage 329de du Protagoras comme source possible de la notion d'homeomere. II est curieux de constater que dans la suite du dialogue de Platon, Socrate confond Protagoras a !'aide d'arguments sophistiques jouant sur la notion de «partief) d'une entite, laquelle conduit a des apories dans le cadre de l'atomisme logique. D'oU la question : Quel fut le statut du «substrati> dans l'Academie ? (5) Rappelons que pour Aristote le continu (auvq�r;) est le «substrat intrinseque» (U7toxdµevov xix0' ixUTO) de l'infini (&7te�pov) III 208a. Or 1'&7te�pov est frequemment (sinon toujours) identifie a l,a matiere {6A"IJ). Cf. la citation I I I , 207a35.

494

HOM:EOMERES hT ANHOMEOMERES D'ARISTOTE

RENE THOM

Pour un ensemble stratifie usuel de la Mathematique (par ex. un ensemble defini par des equations analytiques fi(Xj ) = 0), ii y a toujours une stralificat.ion telle que Jes strates aient des plus tangents qui sur leur bord se recollent continueJYlent6• ll y a des modeles polyedraux de la stratification en tout point d'un tel ensemble ; une strate ne peut se ramifier indefiniment quand elle se rapproche de son bord. La notion d'homeomere souleve d'assez grosses difficultes concep­ tuelles pour qui n'a pas la « culture » que donne !'elude topologique des espaces definis en Geometrie Algebrique ou Analytique. Au sens strict, un homeomere definit - lorsqu'il s'agit d'un homeomere de dimension trois (un <( volume »)1 une nature _qualitative locale de la matiere sensible - pour la science moderne : un etat physico-chimique local. Ainsi pour Aristote, un element tel que l'eau, une substance minerale telle que le cuivre, I' argent, un milieu organique « homogene » tel que le sang, la moelle, l'interieur de l'os, la « chair», sont des homeomeres. Ainsi defini par une qualite phenomenale locale, l'homeomere n'a pas de fo r_me pro pre. Mais il a cependant une « quiddite », un <( logos », qu' Aristote associe a ses capacites d'interaction lorsqu 'il est melange avec d'autres homeomeres. (Mais ce « logos>), dit-il, est difficile a definir : o i. A6yoL ocU't'Wv ou>< &:>}. Voir a ce sujet « Stabilite Structurelle et Morphogenesef>, notion de point regulier et de point catastrophique [SS!\-1 , p. 40]. Les ensembles stratifies sont caracterises techniquement par des propriete-s de regularite locale de leurs plans tangents (Proprietes a, b de Whitney). Pour leur definition, voir le livre recent (< Stratified Morse Theory 1>, par M . Goresky, R. Mc Pherson, Springer Verlag, 1988. =

A

AUJOURD'HUI

495

strates de dimension < 2. Une strate de dimension deux est une surface qui separe deux milieux homeomeres qualitativement distincts ; ainsi les surfaces telles que les membralles pulmonaire ou intestinale sont des homeomeres de dimension 2 (transversalement, ce sont des anhomeome­ res puisqu'elles separent deux milieux differents). Une strate peut avoir une topologie intrinseque qui lui donne une forme globale (non euclidienne). Ainsj Aristote a remarque qu'une veine est topologique­ ment un cylindre S1 X I (avec les notations bourbakistes : S1 = cercle, I = segment, X = produit d'espaces) ; lorsqu'elle est fendue longitudina­ lernenL, elle donne naissance a des fragments, homeomorphes au carre I X I, qui ne sont pas de meme forme que la veine cylindrique initiale (P.A. I I 647a19). Signalons encore une autre difficulte : Une strate, avons-nous dit, ne peut se ramifier indefiniment quand elle se rapproche de son bord. C'est cependant, apparemment, le cas pour certaines structures biologi­ ques : ainsi l'homeomere sang a pour bord le tissu vasculaire, qui constitue la paroi des veines et des arteres. Or veines et arteres se ramifient en des conduits de plus en plus petits, jusqu'a former des capillaires invisibles a l'reil nu. lei encore Aristote a eu conscience de cette difficulte : ii compare le systeme vasculaire aux rigoles d'irrigation d'un jardin (P.A. I I I , 668a l0-13) ; et ii n'hesite pas a dire que le sang n'est pas une « p artie i> de l'organisme animal proprement dite : TO i:x!µi:x : ou3ev y<Xp �0iv �wv µ6piav (P.A. 656b21) : En effet le sang est pratiquement present (partout dense, diraient les mathematiciens) dans tout l'organisme, sauf dans quelques excroissances peripheriques comme pails, dents et angles. Ce n'est que tout recemment que !'attention des scientifiques a ete attiree sur ce type de formes ramifiantes, baptisees « fractales » par B. Mandelbrot. (Voir B. Mandelbrot 1977). II. Genos et Eidos.

La notion d'homeomere souleve une autre difficulte - plus essentielle peut-etre -. En effet, !'equivalence phenomenologique locale ne peut etre definie que si on sait decider si deux voisinages U x , U y de deux points x, y de l'organisme ontou non meme apparence qualitative. Or ici nous rencontrons une notion fondamentale de l'aristotelisme, a savoir, le couple genos-eidos, traduit canoniquement en Genre-espece. Deux qualites (a), (b) (par ex. deux couleurs) sont dites appartenir au meme genre, si on peut les deformer continuement l'une dans l'autre (par experience mentale). Le genre doit done etre vu comme une « UA1J )>, un continuum qualitatif dans lequel les differences (3ioc<popoct) decoupent des frontiCres qui separent ce genre en e:t81J distinctes. On peut ainsi dans une metaphore dynamique voir le genre comme un fleuve se ramifiant daris un delta (chacune des branches de la ramification constitue un eidos). II

496

RENE THOM

est clair des !ors qu'on pourra definir deux modes d'equivalence selon la phenomenologie locale : en imposant aux deux points x, y d'avoir des v01smages rentrant qualitativement soit dans un genre (sensible) . commun, so1t clans une espece commune, la premiere relation etant moins fine que la seconde. On posera en principe qu'un re:\loi; n'est autre qu'une classe d'equivalence definie par l ' isolopie (done l'homeomorphie) des str� tific� t10ns (y) (y') definies sur la boule (B') par !'equivalence phenomenale locale selon le « genre ». Si on adopte !'equivalence locale du point de vue plus fin de I'equivalence des « eide 1>, alors on definira une « espece » (•tso,). De maniere plus precise : deux organismes (0 1) (02) sont d1ts de meme Genos g, s'il existe un homeomorphisme H : (lFl', 0 1) -+ (JR', O,) telle que I'image par H d 'un homeomere (h 1) C 01 de genos r est un homeomere h2 = H(h1 ) de meme genos r7• Comme l'a bien vu °.'Arey Thompso � ce qu'Aristote appelle !'equivalence par exces OU par : . defaul est en fa1t I eqmvalence topologique. L'auteur de « On Growth and Form», qui fut un traducteur de l ' Historia Animalium (et un commenta­ teur s� uve � t original de ces textes) avail sur ce point essentiel compr1s Ar1stote beaucoup mieux que Jes commentateurs traditionnels. ll reconnait avoir tire d' Aristote l'ictee de ses celebres diagrammes ,. d eqmvalence entre especes de poissons8• Appelons G !'ensemble des g�ne �ss� cies a un groupe biologiquement defini (soil « populairement», _ c est� a-d1re 1nformellement comme le revoc; des poissons, soit une « spece » (au �ens moderne) definie par la stabilite de la reproduction � (l mterfert1'.'t�) des md1v1dus de ce groupe) ; pour !'equivalence plus fine selon les e1de (on admettra que ces eide ont une certaine stabilite herectitaire dans ce second cas). La classe d'equivalence sera dite egalement une espece (eidos). Chaque espece est dans un genre. On a done une application canonique 'Y de !'ensemble (•) des eide dans !'ensemble (G) des gene. Ainsi, par ce morphisme, 'Y : (e) -+(G), chaque genos g e G . . ensemble 'Y - 1 (g) d'eide dont les representants se subd1v1se en un sertam . . ont une orgarnsat10n 1dentique (meme « type topologique ») selon l'eidos. JI suffit pour cela qu'au moins un genre I' quali.fiant un homeomere pris . dans le representant canornque du genre (g) admette une variation topologique (ou qualitative) dans sa subdivision en o'l3� . Bien entendu la �

�

(7) L'homeomorphisme II peut ne pas s'etendre a des organes petits, fins et nomb1:eux « Deux chats n ' ont pas necessairement le meme nombre de pails)) (Thom, SSM). ?n do1t au naturaliste Frangois Grandjean la distinction entre caracteres ortholaxiques (lies a la str� cture et conserves par H), p/elhotaxiques (non conserves numeriquement) et cosmolax1ques (non conserves numeriquement, mais dont la plage d'occupation est gl� b�l�me? t c�nser�e en f�rme). V?ir a ce sujet !'article de Yves Bou ligand (! Listes de . prionte )>, Ill B1olog1e Theor1que, Sohgnac, ed. Y. Bouligand, Editions du C�RS 1989. (8) Diagram �es de d ' � rcy Thompson, in « On Growth and Form )), Sir d'Arcy _ W. Thompson, Abridged ed1t1on, ed. J . T. Bonner, Cambridge Univ. Press' 1961 ' pp. 300301, citation d'Aristote p. 273).

E A AUJO URD' HUI 497 HOME OMER ES ET ANHO MEOM ERES D'AR ISTOT souvent qu'il lusoi re, car la precision de cette definition n'est le plus tres theoriquement (La division d'un r n'est vrai, Tout ce qui vien t d 'etre dit de ) n'est nommemenl hora evidemment que si aucune difference (diap entendu, .G�nos et bien s, specifiee. Si Jes differences sont specifies' . alo � ue ces differences lorsq ; fiees eide sont de ce fait automatiquement spee1 entre ovipares et n ndio disti Ja portent sur Ja physiologie (comme dans . ns), alors ou aer1e s t1que aqua , stres vivipares, ou celle entre anim aux terre la definition d'equivalence Ies eid"e peuvent n'avoir rien a faire avec y a des relations entre .Jes morphologique donnee plus haut . En fait, ii , la j ustement, entre en ieu deux definitions (tout oiseau a des ailes), mais structure et fonction. Nous Y toute Ja problematique des rapports entre reviendrons. discutee dans CAA , L'icte e de definir toute difference &v �c;i A6yo;> ologie d'atomisme substan­ pp. 94-1 00, me semble appartenir a Ja myth Tres tlit Ar1stote s'est rendu tialiste qui fut celle du premier Aristote. nnels et qu'on ne pouva1t les compte que Jes genres etaient multidimensio s qu'a u prix de contors10ns engendrer par des couples de contraire Ja etaphys1que donnent une Jogiques dont Jes obscurites du Jivre .r de n a a sa d1spos1t10n aucune assez fidele imag e. Mais eom me Ar1stote es, ii en est reduit au seul genr representation multidimensionnelle des 1onct10n par ha1son des Ja : u moyen de description de Jui conn differences perceptibles des lles ente contraires. Par ailleurs trailer d' accid admissible de la part d'un comme Ja couleur serait difficilement

;-t

(*) Par le sigle CAA nous designons l'ouvrage de P. Pellegrin : La classification des animaux chez Aristote, Les Belles Lettres, Paris, 1 982.

498

RENE THOM

biologiste. Sans doute l'ousia est-elle (Aristote le rappelle en P.A., 644a, 29) un « atome selon la forme » : To Tiji .t8.. lhoµov. Mais l'eidos est alors la « forme propre » de l'ousia, et non un universel general comme la couleur. Dire que : ou 8tocq>opdtv 7tote'i 1J 51'� (Mel. I 1 058b6) (la matiere ne !ait pas · de difference), n'empeche que c'est la matiere sensible qui supporle la difference et qui la manifeste. L'axiome precedent vaut (peut-etre) de la mythique « materia prima )), mais non de la « materia signata » qui est la seule accessible a !'investigation de l'observateur. II est vrai que tout genre est divise par des differences « specifiques ». C'est ici qu'un modele geometrique de la predication peut etre utile : Soit G un genre de qualites sensibles (la couleur, par exemple). Supposons que la matiere du genre' constitue un espace euclidien G (comme l'espace lR s+ des impressions de couleur). Les differences (diaphorai) seront portees par un ensemble catastrophique (K) ferme el canoniquemenl defini dans (G). L'ensemble (K) va decouper dans (G) des blocs distincts (•i) qui definiront Jes eide du genre. L'etat qualitatif d'une entite premiere X relativement au genre (G) (par exemple la couleur) est definie par une application continue a : IXl -+ G (IXI ctesigne le substrat de l'entite X) oil ,,- 1 : (ei) sont les homeomeres de X. C'est la ce qu'Aristote a voulu exprimer par la formule : « la matiere n'opere aucune difference (dans le genre) », formule qu'on comparera a : « La difference est l'eidos dans la matiere », P.A. I , 643a24. Dans ce forma-Jisme, !'ensemble K decrit le genre et sa decomposi­ tion en eide ; !'application a decrit la qualification de la matiere (du ressort de !'accident, selon l'Aristote de Met. J). Mais une moriologie (si elle existe) a precisement pour but de controler ces accidents et d'en donner raison. III. Les Anhomeomeres.

En depit du « flou » inherent a la definition d'une qualite, la notion d'homeomere est une notion conceptuellement precise ; en ce · sens, la partition (la « stratification») d'une organisme en ses homeomeres definit ce qu'en termes modernes· on peut appeler son « organisation ». II s'agit 13. d'une definition intrinseque, morphologique, celle que decrivent les planches anatomiques. Mais les elements de cette partition (Ies strates)

(9) L'emploi du terme « matiere du genre» est considere par certains commentateurs comme une pure metaphore. Mais comme le dit excellemment V. Cherniss : « I f it is a metaphore, it is a metaphore of fundamental significance ». Dans la definition du Genos (Mel .6. 28) comment concevoir !'expression TO 7tp6°l't'ov x.wljaixv 6µoet8kc;, 1024b8, le « mouvant originel indifferencie l), sinon comme la matiere primitive qui va s'ecouler a travers le crible des differences pour se fragmenter en blocs partiels (Jes eide)? (aspect que la traduction de Tricot rend fort peu).

HOMEOMERES ET ANHOMEOMERES D'ARISTOTE A AUJOURD1HUI Cheveux

v F��nt

ane / �

Hie

I

\

\\

Sourcils

y��: /, \--...... I

Paup1eres

' --.. Narmes Face - Nez I -..... Bouche

.... .... ..... ..... .... ....

\

\ \

Cou

,,, .... /

,'

',

Corps humain

Menton

,,, ....

Langue

Dents

Coude

Levres Pouce

,,.,/'

Ma1n . ...... _.,,. ..... ,,,.# ......._ Bras --Avant·B ras ......._ \ Doigts �au/ /'

,,,,,. o1

Thorax

'

Trone

499

�

'

,

,

,

',

Diaphragme ,,, "'

./ / ./ ./

Abdomen ------ Ventre - Nombril

Fig. (C)

./

'� Cuisse - Jambe --- Pied __... Orteils

Hanche -""

..... ..... ....

/,,,;

Genou

Diagramme (M)

ne repondent que difficilement a la notion usuelle de « partie >) d'un animal. Par exemple le « sang» existe pratiquement dans toutes Jes parties usuelles du corps humain (tete, bras, peau, etc.) souvent sous forme invisible ; et Jes strates de dimension 2, I , 0, ne portent en general aucun nom. C'est pour etre en mesure de retrouver les parties au sens usuel - traditionnel - de l'organisme qu'Aristote a introduit le concept d'anhomeomere. Un anhomeomere est une reunion d'homeome­ res. Mais ii n'est nullement precise chez Aristote comment se definit l ' individualite d'un anhomeomere, ni meme s'il a des frontieres morphologiquement definies. La seule indication un peu precise est que l'anhomeomere est le siege de travaux et d'activites (�pyoc "°'' 7tpoc�
RENE THOM

HOMEOMERES'ET ANHOMEOMERES D'ARISTOTE A AUJOURD'HUI

Si !'on considere les parties du corps humain, telles qu'elles sont definies dans le langage nature!, on observera d'abord qu'il y a une tres grande r nniversalite » de langue a langue dans cette decomposition 1 0• En sorte qu'on peut penser que cette partition est fondee sur des · mecanismes psychiques profonds communs a toute l'humanite. (Tres probablement les partitions analogues faites sur le corps des animaux procedent d'une projection anthropocentrique). On peut s'efforcer de decrire cette organisation par une structure ramifiante en arbr�. On constate d'emblee qu'il y a d ifficulte a presenter cette decomposition <( naturelle » par un arbre constitue de ramifications dichotomiques successives. En effet, d'une part certaines structures sont visiblement transitionnelles comme le Cou entre Tete et Trone, l'Epaule entre le Trone et le Bras, la Hanche entre le Trone et la Cuisse, le Menton entre le Cou et la Bouche . . . en sorte qu'on ne peut se contenter d'une structure en arbre. On a besoin a la fois d'une relation d 'inclusion (X partie de Y) (notee X ->- Y) et d'une relation de proximite (A contigu a B) notee A . .. B ; la relation d'inclusion a tres souvent une interpretation fonctionnelle (tel doigt fait partie de la main), alors que celle de contiguite n'a souvent aucune fonction evidente. L'individualite de chaque anhomeomere est done souvent difficile a definir ; en effet, les « organes » au sens traditionnel du mot n'ont souvent pas de frontieres morphologiquement evidentes. Si !'on considere par exemple un couple d'anhomeomeres tel que Pouce -+ Main, on constate qu'il n'y a pas d'homeomere qui definisse la frontiere entre le Pouce et la Paume de la Main. Au niveau du squelette, on peut certes trouver une telle frontiere dans la surface d'arliculalion qui separe la phalange du Pouce d'avec l'os contigu de la paume (le metacarpe). Mais cette « cloison » ne se prolonge ni dans la chair qui entoure les os, ni sur la peau, oU elle n'est marquee que par une abrupte difference d'orientation (dans la zone interdigita­ le11). Ceci montre qu'il n'y a pas coincidence entre la description d'un organisme par homeomeres et celle definie par ses anhomeomeres. Un anhomeomere n'esl pas nCcessairemenl limite par des homeomeres de la siralificalion du « Genos ». On voit qu'ainsi la distinction homeomere-anhomeomere recoupe la definition moderne : Structure-Fonction. Les homeomeres definissenl la structure, Les anhomeomeres Les fonclions. Comme le rapport entre

structure et fonction est un probleme crucial de la theorie biologique (ce fut le theme de la celebre controverse academique de 1 830 entre Cuvier et Geoffroy Saint Hilaire), on congoit !' importance qu'il y a de preciser ce rapport. Pour cela nous evoquerons d'abord un cas pur.

500

(10) Sur l'universalite des denominations des parlies du corps, voir Andersen, ES, 1 978, Lexical Universals in Body parts terminology in Greenberg, ed. 1 978, Universals of human Language, Stanford, tome 3, 335-86. (11) On observera que ces zones digitales, issues de la palette palmee chez l'embryon proviennent par necrose genetiquement programmee de ces secteurs interdigitaux. Evidemment la variabilite angulaire des directions des doigts necessite cette destruction ; les frontieres de l'anhomeomere pouce contiennent une telle zone fonctionnellement variable - mais non diffCrenciee en tant qu'homeomere.

501

Le cas pur. II s'agit du cas oil l'anhomeomCre est compose d'un seul homeomere. Un tel anhomeomere ne peut etre de dimension trois, c'est done au plus une surface, cloison ou membrane entre deux homeomeres distincts et contigus. C'est le cas de certains anhomeomeres cites par Aristote : l'intestin, le poumon. Considerons ce dernier cas. La vocation fonctionnelle du poumon s'exprime dans le cycle d'hysteresis aVAv issu de la catastrophe fronce (voir Thom [1986]).

c

+

'

02

v

u

Sang veineux

I

CO'

Organisme extra pulmonaire

a

t

v

Poumon

Air exterieur

Sang arterial x Diagramme (A)

I

I

I I

VA poumon av organisme extrapulmonaire vV sang veineux Aa sang arteriel en ce cas l'anhomeomere est une surface, oU se font les transferts gazeux entre !'air et la sang. Si !'on paraphrase la terminologie de P.A. I i , 646a l 3 , 646b20, on sera amene a dire que les homeomeres v V et aA sont en puissance (car ils ont des pouvoirs (3uv6qm,) Ibid. 646b20) et ces pouvoirs se realisent en acte sur les verticales av, AV ... En projetant le cycle sur !'axe de la variable de controle X, les organes (Poumon, Chair) sont en acte par

502

RENE THOM

rapport a ces homeomeres mobiles que sont le sang veineux et le sang arteriel. Dans un cas de ce genre, I' organe support de I' acte est « bord » de l'homeomere qui se dirige vers Jui. Situation qu'on resumera par l'axiome ABP : L'acte est bord de la puissance. Dans un article a paraltre (Thom 1 988) j'ai etudie Jes connotations geometriques de !'opposition puissance-acte telles qu'elles apparaissent dans le corpus aristotelicien. La regle ABP s'appuie sur quelques metaphores essentieHes chez Aristote : la matiere (u1'11) s'identifie a l'indefini (ome<pov) ; et la forme (ec3o.;) limite la matiere comme la sphere bord limite le gaz contenu dans un recipient spherique . . . Typique est la phrase (, I I I , 207a35) : l'infini est enveloppe cornme une matiere interne, c 1 est la forme qui l'envelop­ pe : 7tE:ptE:xe:'t'CX.L yOCp We; � Ut..11 e..rrOc; xal. 't'O &7te:tpov, 7tE:ptE:xe:t 8E TO e:I3oc;. La forme, c'est toujours l'enveloppe (7t IV, 212a) (la premiere limite fixe de la theorie du lieu), l'enveloppe de la theorie des enveloppes : Le solide au contraire, rigide, a une forme identique a elle-meme. C'est par essence le domaine du T6 Tl �v ltvcx.t, oil la forme - permanente - s'identifie a son substrat : pas etonnant done, si c 1 est aussi le domaine du Logos ! Le cas intermediaire est celui du liquide (I'eau dans un tonneau) ( IV, 21 l b3) : la la forme de l'eau est delimitee d'une part par la forme du contenant (le tonneau) et d'autre part par la quantile du liquide contenu. II y a done ici une determination mixle de la forme - la matiere intervenant par sa quantile, sa « masse ». Ce dernier type de rapport matiere-forme semble fondamental dans certains mecanismes de l'Embryologie. (C'est celui decrit par la metaphore celebre du moule inlerne de Buffon). De maniere generale !'opposition puissance-acte est affectee d'une essentielle ambigufte. D'une part, on peut dire que du point de vue de !'evolution temporelle, la puissance engendre l'acte. Or, on sait (et les anciens le savaient) que le point par son mouvement engendre la courbe, la courbe par son mouvement engendre la surface, et la surface le volume. D'ou l'idee que ces engendrements successifs conduisent de la nature potentielle, essentiellement privative, du point geometrique a l'actualite parfaile du corps tri-dimensionnel. En ce cas, on dirait que la surface-bord est en puissance par rapport au corps tri-dimensionnel en acte qu'elle limite, contrairement a l'axiome (ABP) (Cf. la discussion de la quatorzieme Aporie in Met B5 et Met M2, 1 077a27-30). Mais en fait un nouveau facteur intervient : en fonction du caractere ouvert ou ferme de l'« interieur», la situation change du tout au tout ; le bord de la statue est forme, ii est acte par rapport a la matiere (informe) de la statue. Par contre ii parait legitime de dire que le bord (la surface) est en situation de puissance par rapport a la statue achevee, car e'en est une partie (µepo.;) . Si l'on tenait a ordonner par ordre d'r nctualite» («0useite») les figures geometriques eiementaires, je crois qu'on devrait les presenter ainsi : en dimension un

r I

I

HOMEOMERES 'ET ANHOMEOMERES D'ARISTOTE A AUJOURD'HUI

503

le segment ferme 0 ;5; X ;5; I est l'ousia pleine et separee le point X = I bord, c'est l'eidos de cette ousia 0 ;5; X < I le segment semi-ouvert J : c'est la matiere. On pourrait dire aussi que le segment semi-ouvert J est « en etat de privation», privation comblee par !'acquisition de son bord. Mais comme l'a observe Aristote la privation n'est pas I' absence de forme ; c'est plutot une « lacune» de la forme ( I I , 193b!9-20) qu'on peut representer par la presence au bord d'un point de fracture virtue! - dont !'actualisation en deux levres briserait l'etancheite de la paroi, et mettrait ainsi l'entite en peril (un homme qui, blesse, perd son sang, est dans un etat de privation catastrophique). Ou encore, dans le cas d'un predateur aflame, c'est un etat de metastabilite qui peut revenir a l'equilibre par la rencontre d'une forme exterieure (la proie). La privation peut etre aussi une sequelle irreversible d'un choc, la « cicatrice » d'une mutilation organique. Ceci nous suggere de discuter le statut de l'axiome (ABP) pour Jes discontinuites organiques (strates de dimension deux, par exemple : membranes ou cloisons). L'idee essentielle est que, dans la description de processus transients, la structuration temporelle du processus physiologique, son deroulement, determine l'opposition puissance-acte ; generalisa·nt l'exemple donne dans la formule : « La route d'Athenes a Thebes est la meme que celle de Thebes a Athenes (, I I I, 3, 202al4) », la meme disposition organique peut recevoir, selon le contexte dynamique oil elle opere, des interpretations differentes de !'opposition puissance-acte. Par contre lorsqu'on 'a affaire a un processus « stationnaire » (un ecoulement permanent - comme I' est la circulation du sang -), la regle (APB) vaut en general. Considerons par exemple le cas des membranes cloisons entre deux types de tissu. Certaines de ces surfaces sont en etat normal le lieu de transfert de matiere ou d'energie. Mais d'autres surfaces apparaissent inertes et ne sont le lieu d'aucun transfert fonctionnel. Par exemple un os comme le tibia est un anhomeomere limite d'une part par des surfaces d'articulation avec les os contigus (la rotule du genou la malleole du talon). La le schema du conflit (de « catastrophe ») defini par l'axiom e (ABP) s'applique sans difficulte. Par contre la surface laterale du tibia couverte par le perioste, qui n'est le support d'aucune activite fonctionnelle evidente, ne peut recevoir une interpretation qu'en remontant a l 'ontogenese : la necessite qu 'a l'embryon de mammifere de marcher lui dicte d'avoir a creer des membres. Seule la dynamique originelle du bourgeon de membre peut expliquer la discontinuite locale de la surface laterale du tibia. D'ou la necessite de revenir a la structure globale de l'etre vivant, done de trailer.

504

RENE THOM

IV. De I'ame ou des ames.

Car le rapport entre structure et fonction souleve un probleme qu'on ne peut pas esquiver. C'est celui de ce qu'Aristote appellera dans le De Anima : l'dme. La formule celebre du De Anima 412b, 18-19 : « Si l'reil etait un animal, son a.me serait la vision », doit etre rapproe.hee d'un modele de l'organogenese de l'ceil comme du a !'impact d'un « rayon de lumiere virtue! » sur l'ectoderme competent. On connait en Topologie un emploi metaphorique du mat ame, comme lorsqu'on dit qu'un tore plein (D2 X S1) : (produit du disque D2 par le cercle S', un anneau) a pour dme le cercle axial (0 x S1), lieu des centres des boules de rayon constant dont l'enveloppe est le tore plein (la surface du tore est l'enveloppe des spheres bards). Je crois qu'il y a la plus qu'une coincidence accidentelle. L'ame - support de l'individualite - est en principe un point. Mais ce point n'existe qu'«en puissance ». Pour se realiser, ii doit s'epaissir et se construire un corps tridimensionnel, et l'organe sera le bord (l'eidos) de cet epaississement. Appelons « link » (avec Jes topologues) le bard d'un tel voisinage tubulaire. Alors l'acte de l'ame en puissance est localisee dans son link. Par exemple si !'unite et l'indivisibilite de l'ame de l'etre vivant est representee par un point dans IRS, alors le link de cette ame sera la peau, membrane qui separe le milieu interieur vivant du milieu exteriellr inanime. Mais ce serait, je crois, p3rfaitement illusoire de chercher a localiser ce point (le siege de l'ame !) dans l'organisme. Revenons au diagramme (il). Si on veut localiser l'ame de la fonction biochimique qui apporte au corps l'energie exterieure (par l'oxydation c + 02 -> C02), alors on doit considerer un genre bidimensionnel, a savoir le plan defini par deux variables : x, y La variable y sous-tend !'opposition dechet de la combustion aliment carburant C02 0' catabolisme anabolisme et la variable x !'opposition Interieur = Exterieur ; le link du point « animal » (representant l'ame) sera le cycle qui, defini a la Van der Pol, donnera le cycle du sang (diagramme il). Ce cycle est initialement un cycle d'entrainement du type « roue de moulin» dans le flux energetique de pyramides ecologiques (voir Thom [I 986]). Cet exemple montre que le « link » d'une ame se forme dans un espace abstrait (espace fonctionnel). A partir de l'etat quiescent de !'ovule avant fecondation (attracteur ponctuel du metabolisme), !'explo­ sion consecutive a la fecondation fait bifurquer cet attracteur ponctuel

l I

I

HOMEOMERES' ET ANHOMEOMERES D'ARISTOTE A AUJOURD'HUI

505

en un attracteur multidimensionnel, oi.t toutes les fonctions organiques universelles sont representees (ce que j 'appelle la blastula physiologique). Puis cette figure se deforme localement en creant des zones catastrophi­ ques qui vont se localiser en differenciations cellulaires, puis en organes spatialement localises. La deformation de cette figure initiale est ctecrite par !'application (<>) qui associe a chaque cellule de l'organisme son etat metabolique local. La deformation "t de !'application " est modulee selon Jes grands gradients morphogenetiques de l'organisme tel que le gradient cephalo-caudal. C'est la variabilite (catastrophique) de cette application " chez l'adulte qui caracterise l'organe. Parfois cette variabilite fonctionnelle n'est pas visible en tant qu'homeomere ; ainsi s'explique le fait que certaines limites d'un anhomeomere ne soient pas constitues d'homeomeres visibles, par exemple Jes zones interdigitales decrites au § 311• Le cerveau est chez l'adulte la projection spatiale de !'unite dynamique originelle du metabolisme de la blastula. Le SNC (systeme nerveux central) innerve tout l'organisme comme le sang irrigue tous les tissus. II faut retirer de tout ceci que l'ame d'une fonction est au mieux realisee par une strate de grande codimension (0, I , 2) ; le plus souvent, elle est localisee dans un espace abstrait comme l'espace de genre, et l'organe correspondant est le « link » de cette strate (c'est le bard d'un epaississement)12. II resterait ensuite a exprimer l'interdependance fonctionnelle des 3.mes organiques. II faut voir la formation successive de ces 3.mes comme une brisure d'une figure initialement continue qui y creerait des aretes ; ainsi nait une stratification d'un espace « fonctionnel » (au sens de la mathematique�. La la dependance fonctionnelle (la <>uvaba aristoteli­ cienne) s'exprime par une relation d' incidence entre Jes strates de cette

(12) L'aspect geometrique de la relation entre structure et fonction est chose delicate. On sait en Topologie ce qu'on appelle la «dualite de Poincaref> . C'est \'operation qui consiste dans une espace de dimension trois a associer a toute cellule de dimension (i) (0 ::;; i ::;; 3) la cellule « orthogonale» de dimension complementaire 3-i. Par exemple dans IR3(x, y, z) , au segment de droite verticale IJ (1(0,0 + 1), J(0,0 - 1), on associera le disque orthogonal z 0, x2 + y2 < 1 . On peut dire « grosso modo » que structure et fonction se realisent en des figures geometriquement duales. Exemple pour i 2 (surface} le bord du dual est la sphere S0 constituee de 2 points : c'est la situation de la membrane separant deux types de tissus. La fonction est le lransferl de materiaux (fluides) d'un milieu a l'autre·defini par le segment transverse (oriente ou non); la structure est la membrane limitante. Autre exemple : !'articulation entre 2 os contigus dependant d'un seul parametre, !'angle plan entre Jes directions des deux os contigus 01, 02• Alors la structure est la surface d'articulation. Une telle surface (cylindrique) est elle-meme duale (partiellement) du centre de la rotation w de 02 par rapport a 01 pris comme repere fixe. Une rotule spherique permet au contraire une articulation a deux degres de liberte. L'«3.me» est alors !e centre de la sphere d'articulation (cf. la tete du femur). =

=

506

RENE THOM

stratification. La « moriologie» s'exprime alors comme des contraintes de nature hereditaire portant sur !'application cr qui envoie le substrat materiel (le corps) dans cet espace fonctionnel. Ainsi se trouvent formellement separees les contraintes universelles de la Physiologie des choix « plus ou mains accidentels)> realises au cours de l'evolution qui conduisent a la localisation des fonctions en des organes diversement situes ou conformes. V. Anhomeomeres et Finalites.

Je voudrais explorer ici le rapport entre la methode « diairetique» fondee sur l'emploi d'un questionnaire 1 et la structure en arbre que nous avons relevee chez les anhomeomeres (diagramme M). Cette structure en arbre, reflete en general le processus de leur engendrement au cours de l'ontogenese. Mais par ailleurs, Jes diverses parties de l'embryon subissent elles aussi, au cours de leur developpe­ ment, des scissions, des jonctions, des transformations topologiques et biochimiques diverses. En ce qui concerne les scissions on peut esperer les interpreter fonctionnellement comme correspondant aux reponses d'un questionnaire. Par exemple en Embryologie Vertebree, on posera d'abord la double question (stade de la gastrulation). Vais-je pouvoir manger? Oui -+ Mfsectoderme Non (pas tout de suite) -> Endoderme Puis - Si Oui om Ectoderme Vais-je pouvoir trouver une proie a manger ? Mesoderme non (pas tout de suite) Oui oui Tissu nerveux non Epiderme (La proie une fois localisee) va1s-1e pouvoir l'attraper ? , Oui Mesoderme -> Somatopleure et membres locomoteurs Non (pas tout de suite, plus tard) -> Tissu vasculaire et splanchnopleure (La proie etant capturee et ingeree) Vais-je pouvoir l'assimiler ? ou1 -+ Intestin Non Si Non solide -> Gros intestin Le dechet non assimilable est-ii liquide -> Reins gazeux -+ Poumon Nole. Les dechets liquides, etant d'origine interne, sont traites par des organes issus du mesoderme (nephrotome).

1

\j

f '

I

HoMEoMEREs' ET ANHOMEoMERES D'ARISTOTE A AUJOURD'HUI

507

Les parentheses manifestent la presence de tissus « inducteurs» issus des reponses precedentes ; elles temoignent de processus de j onction dans I' embryologie. Get exemple est a rapprocher de la critique faite au livre I de P.A. sur la dichotomie platonicienne. Au lieu de chercher a caracteriser un animal par une batterie de questions telles que : L'animal est-ii terrestre OU aerien ? L'animal est-ii sauvage ou apprivoise ? Questions qui portent sur des genres independants et qui peuvent etre donnees dans un ordre arbilraire, ii importe au contraire de viser a obtenir par nos questions une description de l'animal. Par exemple, si l'animal est terrestre, demander : I' animal a-t-il des pattes? Si oui : ces pattes sont-elles solides ou « fendues » ? En ce dernier cas : la patte a-t-elle deux doigts ou plus? etc. II n'est pas exclu qu'on ne puisse aboutir par la methode dichotomique platonicienne a une definition correcte (I'homme bipede sans plumes . . . ), mais une telle definition n'aurait guere de pertinence biologique. II faut, au contraire, organiser une structure en arbre du questionnaire telle que la ramification du questionnaire simule la ramification (onlogenetique) du substrat. Ainsi parviendra-t-on a une definition qui soit en meme temps une description. On s'approchera ainsi de !'ideal, postule dans Met Z, 10, d'une definition dont Les parties sont Les parties memes de l'enlile decrite. Or dans l'arbre des parties du corps, on a par exemple Nez � Face _..-----

Crane Tete ----------.....___ Menton - - -

___.. Y eux

Bouche

Un questionnaire associe serait : la voir? Voulez-vous reperer une proie ? oui (une fois capturee)

-> Yeux

la sentir -> Nez la manger -> Bouche

Le gradient vertical de la face s'identifie a un gradient Lumiere -> Air­ -> Eau -> Terre assez semblable au gradient de densite des elements aristoteliciens : Feu - Air- Eau - Terre ; on voit ici la necessite d'organiser les genres aristoteliciens en genres supfrieurs (ou hypergen­ res), par exemple les verbes de sensation (voir, sentir, entendre, toucher) constituent un hypergenre selon la nature de !'element porteur d'information. (Noter que l'oreille moyenne est d'origine endodermique, comme la muqueuse pulmonaire). La structure de ces « hypergenres » intervenant d �ns l'embryologie n'est pas connue ; mais ils procedent

RENE THOM

HOMEOMERES ET ANHOMEOM:ERES D 'ARISTOTE A AUJOURD1HUI

d'une structure en arbre, du type de l'arbre legendaire attribue a Porphyre, les ramifications etant elles-memes modulees transversale­ ment par des ramifications d 'origine cosmique, comme celle des elements d'Aristote (Ies etats phasiques de la matiere [solide-liquide-gaz] pour les modernes). lei intervient la classification des organes selon l'analogie (xoc� ocvocl.oy(ocv), qui admet elle aussi ses eide (cl. CAA p. 152-153). Deux organes sont analogues s'ils procedent de la meme fonction de la physiologie generale, s'ils realisent le meme « ame physiologique )) universelle (cf. CAA, pp. 152-153). L'ame d'une telle fonction n'est plus un point isole <( stricto sensu » , mais un point dans une configuration actantielle - un chemin transverse a une strate catastrophique selon la theorie des catastrophes. Les Gene qui sont les eide d'un hypergenre apparaissent par division selon le type phasique des substrats mobiles parcourant le chemin (ou une modulation morphologique du schema dynamique lui-meme). Voir a ce sujet CAA, pp. 152-53. II est clair que Jes finalites sont elles-memes organisees en arbre. En vue de realiser un certain but (Bo), ii peut etre necessaire de realiser des buts preparatoires (Bl), (Bi) .. dont Jes effets concourent a la realisation de (B0), soit symboliquement :

appropriees . . . II est certain que le systeme nerveux reste la manifesta­ tion organique de I' unite originelle de I'« i\me » eclatee vers la peripherie. II y a dans cette vision de la matiere du genre qui s'ecoule vers les eid€:13, comme la chair (initialement indifferenciee) de l'ceuf se differencie en specialisations « cellulaires » distinctes - elle-meme analogue a la ramification langagiere des anhomeomeres - une analogie dont (peut­ etre) Aristote fut conscient. Mais alors on ne pouvait plus compter sur la seule scission dichotomique binaire14 pour representer les faits, et justifier une theorie de la quiddite causatrice de toute entite qui filt accessible grace au seul Logos. Les critiques qu 'Aristote adresse, au Livre 1 de P.A., a son ma1tre Platon, sont en fait des critiques qu'il s'adresse a lui-meme. Je voudrais en conclusion presenter quelques reflexions sur la finalite aristotelicienne. On en a fait une objection fondamentale a la science aristotelicienne qui serait ainsi en opposition totale avec la science de notre temps. Je crois qu'on a fait 13 a Aristote un mauvais proces. En effet la finalite aristotelicienne n'est jamais absolue, elle ne parvient jamais avec une totale certitude a son but. C'est une finalite valable W.:; E7tt TO 7toAU, en contexte normal ; combien de fois trouve-t-on sous la plume du maitre la restriction : la fin se realisera, s'il n'y a pas empechement, 11.v µ'ij �• 1:µ"o8(�'1J, I I l 99a8-l l . On sait que dans De Generatione et Corruptione, Aristote va jusqu'a dire que la finalite n'agit pas par elle-meme : To 8' ou gv•x� ou "°''IJ�•x6v (324b, 14) ; personnelle­ ment, je crois qu'il est difficile de faire une theorie de l'embryologie sans evoquer une certaine causalite formelle, s'exprimant sous forme de la presence de champs morphogenetiques. C'est dire que le milieu vivant va presenter des entites locales qui, a un certain moment critique du developpement (h' ocpx.�v fov) ( V I I I , 7, 26la l4) vont entrer en privation ; elles chercheront a pallier cecte privation en essayant de trouver dans leur environnement une forme pregnante (comme la forme de la proie pour le predateur affame) qui va declencher en elle le

508

B' B B l ----

!�o

.

(r)

et ceci pourra s'iterer, donnant lieu ainsi a un graphe (r) dont le but ultime est la perpetuation de l'espece animale envisagee. (Aristote parle de auvoc(noc pour decrire de tels moyens auxiliaires). Toute l'embryologie peut etre consideree comme la remontee de ce graphe a partir du but ultime (Bo) - l'etat de l'ceuf apres fecondation - vers la totalite des buts preparatoires extremes *. L'axiome frequent chez Aristote, '!1j yevfoec 5anpov, '!1j uaec 7tp6npov ( V I I I , 26la l3-14) : premier selon . la nature, dernier selon la generation, exprime simplement que plus une fonction est biologiquement fondamentale, plus elle necessite de fonctions et d'organes preparatoires, plus, par consequent, elle se realise tardivement. Dans cette maniere de voir il existe un isomorphisme entre le graphe du questionnaire fonctionnel qui interprete Jes grandes scissions de l'embryologie, et Jes scissions de l'embryon lui-meme en ses grandes .differenciations (ectoderme, mesoderme, endoderme) : la le questionnai­ re ne met en jeu que la physiologie. Par contre des qu'interviennent les elements non propres a !'Ego, par exemple des contraintes dues aux elements (Air, Eau . . .) du monde exterieur, alors les scissions correspon­ dantes creent des differenciations, et par la suite des jonctions (*) Une fois les scissions realisees, on a une periode ulterieure de jonctions realisant Jes connexions organiques fonctionnellements necessaires, periode de redescente vers 80.

509

(13) Assez curieu!.'ement, Aristote associe dans P.A. I I , 647a3-5 les homeomeres aux fonctions de la sensibilite. Ceci n'est pas absurde si on veut bien assimiler la retine de l'ceil, par exemple, a une surface de « contact» entre la chair et le diaphane. Aristote dit d'ailleurs que la vision est une partie de l'reil (lfl IV, 2I lb2) . (14) Dans le De Generatione et Corruptione I I , 330a30-35 Aristote deflnit les quatre elements par un procede de dichotomie binaire portant sur une materia prime : Humide Sec Eau Terre Froid Air Feu Chaud II se felicite de cette trouvaille ; non sans raison, car ii faudra attendre le cercle de Prague et R. Jakobson pour trouver, avec la phonologie (le phoneme defini par ses traits differentiels) une construction analogue - celle qui devait engendrer le structuralisme moderne. On peut se demander si cette decouverte est anterieure ou non au livre I de P.A. ,

510

HOMEOMERES ET ANHOMEOMERES D'ARISTOTE A AUJOURD'HUI

RENE THOM

mouvement comblant cette privation. On peut d'ailleurs envisager ce processus sous forme globale, affectant un certain ensemble de ces entites (le plus souvent des cellules). 11 y a lieu de penser qu'alors - en contexte normal - le processus va s'effectuer suivant une dynamique · structurellement stable (compte tenu des contraintes devant satisfaire Jes privations locales). Ainsi obtiendra-t-on un champ morphogenetique se deployant dans un certain ouvert Ui de l'espace-temps, ou la morphogenese se deroulera conformement a une carte locale sus9eptible de subir des deformations quantitatives (mais non qualitatives !) qui peuvent titre importantes (phenomenes dits de regulation). L'ontogenese ne sera plus alors qu'une concatenation de tels champs morphogeneti­ ques, aboutissant a la fin a la construction de l'organisme acheve. Dans une telle vision, le seul element quelque peu suspect est la nature physique (ou chimique ?) des coordonnees constituant Jes cartes locales des champs. 11 ne fait guere de doute qu' Aris tote acceptait !'existence d'agents « translocaux» susceptibles de transmettre des effets « il distance », i.e. a une vitesse tres grande. Tel est le cas de la lumiere, propagee par cet element non local, le diaphane7• L'hypothese du « Contact » ( IV 9-10) necessaire a !'interaction entre deux entites, n'est guere restrictive, s'il s'agit du « contact)> entre une matiere sensible etendue localement, et une « matiere » sub tile, invisible et translocale s'etendant extremement loin. Aristote, dans son argumentation contre !'existence du vide, observe (avec raison) que la ou on dit qu'il y a le vide, la lumiere se propage ( IV, 214a l 0). Ce probleme, qui fut celui de !'ether au XIX' siecle, n'a ete resolu que formellement par la Physique Moderne, qui postule des symetries globales de l ' Univers - sans specifier la nature des agents physiques qui realisent ou supportent cette symetrie. D'ailleurs Aristote a manifeste son embarras pour expliquer la phora d'un corps vers son lieu nature! (qui peut en etre eloigne) ( V I I I , 255a-b). Tout ceci fait qu'il n'y a pas plus d e magie clans !'existence d'un champ morphogenetique en Embryologie que clans !'existence d'une Joi malhematique d'evolution en Physique. Enfin pour Jui le Nous patheticos et le Nous poietikos decrits au De Anima I I I 5 sont des agents translocaux, ni localises ni personalises ; ce dernier est compare a la lumiere, qui rend visibles en acte les couleurs qui sans lui resteront invisibles. Car on ne saurait imaginer une Physique sans <( entite exploratrice » susceptible de contr6ler son environnement - comme le repere cartesien presuppose un appareillage de mesure des evenements qui se passent dans ce repere, et comme l'animal vigile a une representation permanente de son environnement.

511

REFERENCES

D'ARCY THOMSON, On Growth and Form, Abridged Ed. by John Tyler Bonner, Cambridge Univ. Press, 1961. V. CHERNrss, Aristotle's Criticism of Plato and the Academy, John Hopkins Press, 1930. B. MANDELBROT, The fractal Geometry of Nalure, Freeman and CO., San Francisco, 1982. P. PELLEGRIN [CAA], La classification des Animaux chez Aristole, Les Belles Lettres, Paris R.

1982. 1970, Les Mathematiques Modernes : une erreur pedagogique et philosophique l'A.ge de la Science, Juillet-Septembre 1970, vol. I I I, n° 3, Dunod, Paris, pp. 225242. Appendice p. 235-42. , 1972 [SSM], Stabilite slruclurelle et Morphogenese, InterCditions, Paris. , 1986, Organs and Tools, a common theory of Morphogenesis in Complexity Language and Life, Biomalh. Science n° 16, Springer Verlag, Berlin, pp. 197-230. , 1988, Les intuitions topologiques primordiales de l'aristotelisme, a paraitre dans la Revue Thomiste, Juillet-Septembre 1988, XCVI" annee, T. LXXXVIII, n° 3 : 393409.

Tno M ,

INDEX DES PASSAGES D'ARISTOTE CITES * Categories 2al4 : 91 Premiers analyliques 24a10 24a22 24bl 8 25a28 30al 5 31b31 32a18 32bl 5 34a5 34a34 34a37 34b 19 43a20 46a3 46al7 46a24 53al 7 70a8 70a38 70bl

: : : : : : : : : : : : : : : : : : : :

357 n 259 207 n 352n 350 n 333 n 325n 353 n 324 n 334 n 331 n 332, 333 284 n 284 n 192, 193, 196, 337 156 n 356n 301 301 302

Seconds analytiques 7la8 71a21 7 l b9 72a1 74b21 74b26 75al 75b37

: : : : : : : :

361 n 371 n 354 n 234 228 355 n, 360 n 360n 277 n

76a6 76a37 76bl 1 78a22 78a26 78a28 78a34 78a36 78b35 79a2 8 l a40 8 l b5 81bl8 84bl9 84b22 84b31 88a36 89b37 90a6 90a8 90a9 90a32 9lbl5 92b5 93a21 93a23 93a25 93a30 93a35 93a36 93b7 93bl 0 93b29 93b30 93b37

: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :

418 215 215 91, 149 342 149 193, 336 149n 337 341 374n 336, 371 n 207, 218 284 n 339 339 279 n 148 148, 207 171 182 182 371 n 146 151, 221 182 151 n 149 n 148 147 147 149n 146, 146n, 176 146 n, 221 n 146 n

* Cet index n e recense que Jes citations a u n endroit prilcis, e t exclut done les renvois a des livres ou a des chapitres en general (par exemple Seconds analyliques I I , 1 1). Quand un passage entier est cite, ii est repertorie a sa ligne de depart (par exemple Seconds analyliques I I, 1 1 , 94a20-94b8 sera repertorie a 94a20). Les renvois sont aux pages de cet ouvrage, le « n >) in?ique que la reference se trouve en note.

INDEX DES PASSAGES CITES

514 94a7 94al 1 95a6 96b25 96b26 96b31 96b36 97a23 97a24 97bl4 98a2 98a3 98a7 98a13 98a35 100a6 : 100a7 100a15

: 221 : 146n : 344 : 151 : 153 : 151 n : 153 : 151 : 143 : 166 : 152 : 151 : 152 : 152, 180 : 31 n 193, 418 : 418 : 419

Topiques 100a18 : 189, 192, 196, 239, 250, 252 100a22 : 189 n 100a23 : 188 n 100a25 : 371 n 100a29 : 267 n, 278 n, 282 n 100bl 9 : 191 100b21 : 189, 208, 210, 245, 247, 249, 253, 267 n, 289 101a17 : 336 101a18 : 239, 242 101a l 9 : 242 101a21 : 197 101a25 : 189 n 1 0 la26 : 188 n, 256, 264 n 101 a27 : 188 101 a28 : 198 101a29 : 257 101a30 : 188, 278 n, 280, 286n 101a31 : 217 101a34 : 188, 195, 213, 284 n 101a35 : 195, 198 101a36 : 188, 195, 198, 213, 262, 286, 336 I O l b l : 195 101b2 : 257 1 0 l b3 : 213, 284 n 101bl3 : 246 101b28 : 246 101b32 : 246 101b36 : 247 103bl : 371 n 103b39 : 246 104a4 : 247 104a7 : 246 104a8 : 209, 210n, 2 1 1 , 246, 248, 249, 253, 266 n, 267n

104a9 : 246 104a l 0 : 249 104a 1 1 : 231 n, 248 104a33 : 266 n 104b4 : 245 104b8 : 247 104bl 9 : 209, 209 n, 210, 210n, 249, 250 104b20 : 209 104b24 : 209 n 104b29 : 250 104b31 : 209, 210n 104b33 : 210 104b34 : 250 " 105a4 : 246 105a7 : 270 n 105a9 : 259 105al 0 : 371 n 105a l 6 : 188 n 105a19 : 189 105a35 : 267 n 105b3 : 280n 105b8 : 259 105bl 0 : 259 105bl 3 : 259 105b30 : 218, 283n 108bl I : 371 n 1 1 3 b l 7 : 371 n l 17b2 : 280n 1 17bl 0 : 280 n 1 1 7b28 , 28o n l 19al : 280n 1 19a38 : 281 n 1 19bl5 : 28l n 122a l 9 : 371 n 123b7 : 37l n 126b27 : 202 129b9 : 208 n 1 4 l b5 : 207 1 4 1 b l 3 : 212 141b25 : 193n 142a2 : 234 155b3 : 258 155b7 : 283 n 155bl 0 : 259 155bl 8 : 258 155b20 : 188 n 155b35 : 188 n 155b37 : 203, 208 156a3 : 203 156a4 : 371 n 156a7 : 188n 156a l l : 188n 156a38 : 204 156bl 8 : 204 156b27 : 188n

INDEX DES PASSAGES CITES

I I

157a3 : 188 n 157a7 : 188n 157a18 : l88n, 371 n 157a34 : 203 157b9 : 204 157b27 : 188 n 157b29 : 189n 157b31 : 203 158al : 203 n, 208 158a3 : 203 158a7 : 267 n 158a14 : 189n 158a31 : 270 n, 285n 158a38 : 189 n 158b24 : 285 n 159a3 : 267 159a4 : 204, 204 n, 272n 159al 0 : 200 159a l l : 258, 260, 268 n 159a12 : 200 159a l3 : 259 159a15 : 258 159a 16 : 258 159a25 : 213, 260, 261 159a26 : 261, 268 n 159a27 : 261 159a28 : 261, 272n 159a28 : 272n l 59a30 : 277 n 159a32 : 205, 260, 268 n 159a33 : 200, 256, 258, 260, 261 159a36 : 260 l 59a38 : 282 n 159b2 : 282 n 159b6 : 268 n 159b8 : 204, 205 159bl 3 : 201, 205, 268 159b l 6 : 269n 159b25 : 206 160a12 : 189n, 201, 206 160a l 4 : 201, 268 160a35 : 188n 160a38 : 203 160b2 : 204, 267 n 160b3 : 203 160b5 : 202, 204 160b6 : 204 160b23 : 202 161a24 : 268 n, 272n 161 a29 : 278 n 16lb4 : 200n 161b6 : 202 16lb7 : 202 16lbl9 : 197, 201, 218

161b21 161b28 16lb34 161b38 162a8 162b l 4 162bl6 162b31 163a29 l 63a36 163b3 163b9 163b23 164al2 164bl 164b3 164b8

515

: 202, 218 : 201 : 201, 232 : 201 : 202 : 218 : 218 : 283 n : 198 : 283 n : 198, 280n : 198, 213 : 280n : 371 n : 280n : 280 n : 189 n, 277 n

Refutations sophistiques 165al9 : 2 1 1 165a24 : 277 n I 65a38 : 255, 272 n 165bl : 204 n, 259, 268 n, 272n 165b2 : 271 n 165b3 : 214, 252, 255, 271 n, 278 n 165b4 : 214 165b5 : 213 165b9 : 272 n 166a23 : 354 n 168b4 : 276 16%20 : 276 169b25 : 214, 255, 270 n 169b30 : 267 n 170a20 : 216 170a31 : 276 n 170a34 : 216, 218n, 279n, 280n l 70a36 : 216 170a38 : 216 170b8 : 214, 277 n, 279 n, 280n l 7 la38 : 273 n 1 7 l b3 : 273 n 171b4 : 214n, 218n, 255, 270 n, 273 171b6 : 214, 279n, 280n 1 7 l b9 : 214n, 281 n 1 7 l b l I : 276 n 1 7 l b l 2 : 277 n 172a2 : 215, 277 n 172a4 : 216 l 72a8 : 215, 279 n 172a l 5 : 259 172a21 : 214 n, 274 172a22 : 219 l 72a23 : 219, 233 l 72a25 : 274 n

172a26 : 214 172a27 : 215, 218, 218n, 279 172a29 : 215, 233, 279 n, 280n 172a31 : 278 n 172a32 : 281 n 172a33 : 216 172a36 : 215, 217 172a37 : 215, 217 172a38 : 217 172bl : 217 175a5 : 277 n 175a31 : 283 n 175b33 : 267 n 182b37 : 208 183a8 : 252 183a27 : 272 n 183a34 : 254 183a37 : 199, 239, 250, 252, 270n, 271 n, 281 n, 282n 183a38 : 250 183a39 : 200, 252, 255, 281 n 183b2 : 281 n, 282 n 183b3 : 280n, 281 n 183b6 : 252 183b8 : 254 183bl3 : 254, 282n 183b15 : 254 Physique 184a!O 184a16 184a18 184bl2 192bl 3 192bl 5 192bl 8 I92b33 193a9 193a20 193bl9 198a24 198bl8 I 98b32 199a8 199bl 8 199b26 199b34 202a 14 207a35 2lla7 2 l lb2 2 l l b3 213b23 214a !O

INDEX DES PASSAGES CITES

INDEX DES PASSAGES CITES

516

: 208 : 233 : 193n : 419 : 395n : 396 n : 400 n : 395 n : 395n : 395n : 503 : 29n : 41 : 393 n : 509 : 231 : 231, 398n : 324 n, 438 : 503 : 493 n, 502 : 195 : 509 : 502 : 475 n : 510

247b19 : 419 261a13 : 508 261a14 : 509 Du ciel 270bl6 270b24 280b20 281a7 281al5 281a28 281b2 281b6 281 b 1 5 281 b20 28lb27 284b5 285a29 286a4 292a l 5 302b4 303a20 306a3 306a13

: 479n : 482 n : 328 n : 324 : 326n : 322n : 324 n : 324 n : 324n : 328n : 328 : 474n : 474 n : 341 : 342 : 482n : 191 , 195 : 191 , 194, 196 : 236

De la generation et de la corruption 316a5 : 316a l l 319b8 321 b20 324a24 324b3 329a24 329b7 332a34 336b!O 338a14

194 : 277 n : 222n : 90 : 76 : 77 : 432 : 30 n : 432 : 485n : 346

Meteorologiques 339b22 339b28 343b8 344a2 344a5 344a34 344b19 344b28 345a6 347a28 347a33 347b34 347b36 348al 8 348a21

: : : : : : : : : : : : : : :

482 n 289 191 n 293 320 296 305 304 296 303 303 291 293 294 294

348a33 348b26 349a27 349a28 349bl6 353b28 354a27 355a1 3 355a26 357b24 365a34 366b31 367a15 367b20 369a8 369bl 4 369b28 379b!O 380al 380al l 380bl 3 380b28 382b30 388b!O 390a14

: 293 : 297 : 289 : 289 : 290 : 290 : 303 : 291 : 291 : 319 : 305 : 306 : 306 : 468n : 306 : 482 n : 290 : 30n : 30n : 30n : 30n : 30n : 31 n : 31 n : 494

De l'dme 402bl 6 : 402b21 : 403a25 : 403a28 : 403a29 : 403a30 : 403b2 : 403b4 : 403b7 : 403b8 : 403bl 1 : 404al : 405a25 : 407b21 : 4 1 1 a7 : 412a16 : 412a28 : 412b3 : 412b6 : 414a24 : 415a7 : 416a4 : 418b6 : 427b6 : 428a22 : 428b1 8 : 433a9 :

192 196 116 116 116 1 15, 132 116 1 16 116 116 116 482n 116 421 481 n 89 436 474 n 129 421 406 n 474 n 481 406 n, 415 406 n 236 417

433a!O 433al l 433b3 434a12 435bl 9

: : : : :

517

407 406 n 407 n 407 n 417

De la sensation et des sensibles 436b12 : 416 De la memoire et de la reminiscence 449b9 : 418 451b!O : 420 Des reves 460bl l : 231 De la divination dans le sommeil 462bl 4 : 234 De La Longeuite et de la brievete de la vie 467b2 : 474 n De la jeunesse et de la vieillesse 468a10 : 474n 469b8 : 483 469bl2 : 222 470a23 : 483 475a9 : 483 Du souf[T,e 483a9 : 368 n Hisloire des animaux 486a1 4 : 486a21 : 486a22 : 487a18 : 487a22 : 487a23 : 487a25 : 487b3 : 487b15 : 487bl 6 : 488a4 : 488a5 : 488a7 : 488b12 : 488bl 5 : 488b20 : 488b24 : 488b25 : 488b27 : 489b24 : 488b25 :

412 413 173 162 157 162 157 162 157 157 425 n 427 n 423 411 406 n 407 n 411 411 411 159, 160, 161 n 163

518 488b26 490al 490a6 490a9 490a12 490b7 490bl5 490b19 490b29 490b21 490b24 490b32 491a5 491a7 491a8 491a9 491al0 491a13 491a20 494a27 501 b20 503b29 503b35 504a13 504a20 504 b 1 1 504bl4 504b27 504b28 504b32 504b33 504b34 504b36 505a1 505a8 505al0 505a16 505a27 505b2 505b3 505b6 505b29 505b32 505b35 506a2 506a6 506b4 506bl3 505b26 505b32 507a34 508bl4 508bl7 509b3

INDEX DES PASSAGES CITES

INDEX DES PASSAGES CITES : 161 n : 157 157 : 157 : 157 : 156 : 157 : 157 : 157 : 157 : 1 58 n : 157 : 155 : 176, 194 : 155 n : 340 : 1 56 n : 31 n : 1 15 : 474 n : 471 n : 178 : 178 : 178 : 178 : 178 : 178 : 159 : 179 : 161 n : 171 : 159 : 159 : 159 : 159 : 159 : 161 n : 161 n : 160, 160 n, 164 n : 179 : 158n : 159 : 158 : 158 : 159 : 159 : 160 : 161 n : 158 : 159 : 158, 1 59, 342 : 152n, 1 60, 164 n : 161 n : 161 ·

509b6 510al3 51 la4 5 1 1 a5 513a8 517b2 517b4 520b27 521 b22 52lb25 531 b21 532a13 533a l 533a30 536a4 537al 538a21 538a29 540b5 543a5 543a7 543a8 543a29 543bl 543b23 456bl0 551a4 557b32 560b21 563b29 564a ll 564bl 5 567b1 1 567b20 569a4 578bl6 582bl 583b3 588a I 588a 1 6 588a l9 588a24 588a25 588a26 588a27 588a28 588a30 589al 589a2 589a33 589b5 589b7 589bl2 589bl3

: : : :

·

: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :

158 158 158 160 191 158 158 158 158 158, 161 180 177 1 59 n 161 159 n 161 159n, 161 n 160 160 161 n 161 n 161 n 424 161 n 423 n 427 480n 159n 425 425 425 n 160 425 161 n 424 419 486 n 472 n 409 4!0 4 1 1 , 412 406 n 4 1 1 , 412 412 4 1 1 , 412 412, 414 412 406 n, 427 422 1 62 1 62 162 163n 52, 162

589bl5 : 163 590al3 : 1 62 590a1 4 : 163, 163 n 590al8 : 161 n 590bl : 161 n 590bl0 : 161 n 590b30 : 161 n 59lal : 161 n 59la6 : 161, 161 n 59lal0 : 161 n 591 a 1 1 : 161 n 59lal2 : 161 n 591a13 : 161 n 591al4 : 161 n, 162 591a15 : 161 n, 162 591al8 : 161 n 591a24 : 162 591b14 : 161 n 59lb19 : 161 n 592a1 5 : 62n 596b20 : 422 596b21 : 161 598a l0 : 162 598al2 : 1 62 598al3 : 162 598a l4 : 162 602b2 : 424 608a1 1 : 412 608a1 3 : 4 1 1 , 413, 414 608al7 : 409n, 416 610a22 : 406 n 610b22 : 406 n 6 l lal5 : 409n, 419 61 lal6 : 406n 612a3 : 406 n 612bl : 406 n 612bl8 : 407 612b20 : 406n, 426 612b21 : 426 613b6 : 425 613b15 : 425 613b23 : 407 n, 425 613b25 : 425 614b34 : 407 n 615al8 : 407 n 615al9 : 407 n 616a4 : 407 n, 426 616bl0 : 407 n 616bll : 407 n 616b13 : 407n, 423 616b20 : 407 n 616b22 : 406 n 616b27 : 407 n 616b30 : 406n

618a25 : 425 619a27 : 427 n 619b23 : 407n 619b27 : 427 n 620a21 : 407 n 620a29 : 427 620bl0 : 407 n 62la20 : 424 62lb28 : 416 62lb29 : 407 n, 416 422bl9 : 407 n 622b19 : 407n 622b24 : 407 n 622b26 : 407 n, 426 623a8 : 407 n 623b5 : 181 623bl7 : 426 623b26 : 407 n 624b31 : 407 n 625bl8 : 426 627a6 : 407 n 627a9 : 407 n 627al9 : 426 627bl0 : 415 630b7 : 425 n 630bl5 : 425 630b21 : 406 n, 409 n, 427 630b31 : 428 631a27 : 407, 416 633a29 : 177 Parties des animaux 639b7 : 340 640a6 : 31 n, 346 640a8 : 346 640a14 : 338, 340 640a18 : 25 642a3 : 345 642a31 : 345 642bl0 : 153 642bl4 : 172 642b21 : 153 642b32 : 12 643al 6 : 12 643a24 : 497, 498 643a27 : 153 643a35 : 153 643b9 : 153, 340 643bl0 : 12, 172 644al0 : 12 644al5 : 155 644a1 7 : 172 644a29 : 154 644bl : 154

51\l

520 644b2 : 644b4 : 644bl 0 : 644bl 3 : 644bl 6 : 644bl 9 : 644b35 : 645bl : 645bl 6 : 645bl 9 : 645b33 : 645b35 : 646a6 : 646a8 : 646a 10 : 646al 3 : 646bl 2 : 646b20 : 647a3 : 647al 9 : 648a2 : 648a4 : 648a21 : 649al 3 : 649a27 : 649a30 : 649b8 : 649b9 : 649b21 : 649b22 : 649b25 : 650a2 : 650bl4 : 650b30 : 650b33 : 65la20 : 65l b20 : 65l b28 : 652a9 : 652b7 : 653a27 : 653bl : 656a l l : 656b21 : 658a23 : 660a 14 : 663a43 : 666a l l : 666b21 : 667a l 0 : 668a l 0 : 669a l 8 : 669b5 : 669b8 :

INDEX DES PASSAGES CITES 173 157 155 155 155 155 155 155, 156 n 155 155 159 155 406 n 155, 406 n 406 n 501 499 501 509 495 414 406 n 29 n 222 482 31 n 29n 29 n 222 45 44 222 406n 425 n 414 30n 30n 30n 30n 483n 475 n 471 n 474n 495 424 n 410 425 n 472n 471 n 406n 495 418n 476 n 164n, 181, 181 n

669b31 672b22 674b7 679a4 679a9 680a3 l 683b20 686b26 686b35 687a6 687a15 689bl l

: : : : : : : : : : : :

342 474 n 343 416 416 486 n 474 n 477 n 474 n 478 n 424 n 477n

Mouvemenl des animaux 703a5 : 484 n 703a l 2 : 472n 703a29 : 489 n Marche des animaux 704b9 ; 340 705a28 : 474 n Generation des animaux 715a3 715a5 71 5a22 715a27 716a5 716a13 -716al6 716a23 719a32 719b2 72la2 72la8 72l b l l 72l b l 2 72lbl 4 72l b l 7 72l b20 72lb21 72lb28 721b29 722a2 722a5 722a7 722a l 9 722b6 722bl l 722bl 3 722bl 7 722b30 723a29 723b3

: : : : : : : : : : : : : ; : ; : ; ; : : : : : ; ; : ; ; : ;

21 n 21 103 103 106 106 483 n 106 30n 30n 103 103 19n 79 19n 80, 104 19n 104 104 1 9 n , 80 104 19n, 81 104 80 107 108 109 107 108 22n 80, l 03

INDEX DES PASSAGES CITES 723b4 : 103 723b29 : 22n 723b30 : 8 1 724a3 : 80, 104 725b3 : 106 726b9 : 105 726b30 : 107 727 a l 7 : 69 n 727a25 : 108 727a34 : 30n 727b l l : 22n 728al 8 : 57, 108 728a26 : 57, 106 728a29 : 69n 729a l 0 : 472n 729a l l : 31 n 729al6 : 59n 729al 7 : 59n 729bl 6 : 22n 730b4 : 61 730b5 : 22n 730b l l : 22n 730bl 2 ; 97 730bl 9 : 22n 73l a24 : 424 73lb21 : 21 n 73lb24 : 490 n 732a25 : 103 732b2 ; 424 732bl 5 : 341 733al 7 : 424 733a32 : 103 733b7 ; 424 733b20 ; 103 734b25 ; 437 734b36 : 22n 735a20 : 103 736b2 : 95, 95 n 736b30 : 477, 479, 479 n, 480 737a l l : 97 737al 3 : 22n 737a 18 : 104, 108 737a20 : 57 737a22 : 67 737a23 : 58 737a27 : 57, 106 738a20 : 486 n 738a37 : 59 738b3 ; 67 738bl2 ; 61 738b27 ; 103 739b24 ; 97 739b33 : 95 n 740a21 : 95n

740bl 8 ; 67 74la23 ; 106 74lb33 : 474n 742al 6 : 191 n 742b23 : 31 n 742b28 : 31 n 743a26 : 59 n 743a28 : 22n 744bl : 30n 744b32 ; 58 746bl : 425 n 746b26 ; 57 747a35 : 107 747b3 ; 107 749a34 : 424 749b24 ; 426 75la20 : 425 n 753a7 : 409, 426, 427 753a 1 1 : 406 n 753al 8 : 30n 755al l : 424 756b28 : 30n 758b23 ; 103 759a36 : 426 n 759b6 : 426 n 760a4 : 426 760b27 ; 192, 194 762al 9 : 481 n 762a20 : 481 762bl 4 ; 482 763b25 ; 60 764b3 ; 107 764bl 5 ; 107 765b6 ; 60 765b8 ; 108 765bl l ; 59 766a3 : 22n 766al 0 : 60 766al 4 : 60 766al 5 : 61, 76 766al 6 : 62n 766al 8 : 61, 76 766al 9 : 63, 67, 68 766a20 : 61 766b7 ; 105 766bl 2 ; 61, 101 766bl 5 ; 23, 61, 62 767al : 486 n 767al3 : 22n, 59n 767a15 : 67 767al 7 : 59n 767al 9 : 61 767a28 : 23 n 767a30 : 23 n, 163n

521

522

INDEX DES PASSAGES CITES

INDEX DES PASSAGES CITES

767a36 : 60 767b IO : 6S, 6S n 767bl5 : 23, 67, 6S, 69n 767bl7 : 69, 69n 767blS : 62 767b21 : 4S4 n 767b22 : 62, 63 767b23 : 69, 9S 767b24 : IS n 767b25 : 19n, I02 767b26 : SI 767b28 : 19 n 767b29 : 9S, I02 767b33 : I02 767b34 : 102 767b35 : 62, 66, 69, 70, 70 n 767b36 : 70 767b37 : 64, 69 76Sal : ! S n , 1 9 n 76Sa2 : 63, 70, 9S 76Sa5 : 19n, 9S 76Sa9 : 70 76Sal l : 64, 65, 66, 67, 70, 70n, 71 76Sal2 : 62 76Sal3 : 67 76Sal4 : !Sn, 62n, 64, 70, 73, 99 76Sal5 : 64 76Sal6 : 6 1 , 64, 69, 77 76SalS : 66, 70n 76n, 99, IOI 76Sal9 : 67 76Sa21 : l S n , 64n, 65, 99, IOI 76Sa24 : 19n, 63, 9S 76Sa31 : 99 76Sa32 : 64 76Sa34 : 6 1 , 9S 76Sbl : 74, 9S 76Sb4 : 70 n 76Sb5 : 99 76SbS : 69 76Sb9 : 64 76Sbl0 : 100 76Sbl2 : 100 76Sbl3 : l 9 n 76Sbl4 : I02 76Sbl5 : lSn, 7 1 , 76, 100, I O I , 105 76Sbl6 : 71 76Sb20 : 100 76Sb23 : 76 76Sb25 : 22 n, 7 1 , 76, 100 76Sb30 : 100 769al : 60, 100 769b21 : 61 772al0 : 59n 772al l : 59n

772al 7 772a31 772b35 775al5 775a l7 776a20 776a31 776b33 777bl 6 777b27 777b28 77Sal7 77Sa!S 77Sa22 77Sa32 77Sa35 77Sb6 77Sbl6 779a2 779a34 779bl2 7SOa22 7S0b6 7SOblO 7SOb22 7SOb23 7S2b33 7S2b34 7S3al5 7S4a24 7S4b4 7S6a3 7S6a34 7SSal7

: 22n : 61 : so : IOS : 30n : 30n : 302 : 30 n : 4S5n : 22n, 59n : 4S5 : 20n : 20n : 51 : 346 : 21 n, S2 : 25 : 21 : 51 : S2 : S2 : 51 : 30n : 51 : 51 : 51 : S2 : 23n : 23n : 51 : 51 : 51 : 23 n : 23n

Problemes 942a22 : 4S6 n Metaphysique 9SOb27 9Sla24 9S2bl I 9S3al5 9S5a 16 9S5bl 9S7b29 993a30 995aS 995b20 995b23 996b26 997a7 1001b27 1002alS

: : : : : : : : : : : : : : :

406 n, 410 211 211 223 223 n 3S3 n 263 n 229 367 n 2S4 n 192n 215 370 n 442n 442n

1002a20 : 442n 1004a2 : 195 1004b2 : 281 n 1004bl 7 : 277 n 1004b22 : 214 n 1004b25 : !S5 1007a20 : 436 10!6a32 : 51 1016bl7 : 3S4 1020a23 : 125 1024bS : 439, 49S n 1025a6 : 3Sl n 1025a 10 : 3S3 n 1025a30 : 370 n 1025b30 : 52 1026a 18 : 195 1028a I I : 432 1028bS : 1 1 , 12, 13, 42 102Sb!5 : 12 102Sb31 : 12 I028b34 : 431 1029a2 : 431 I029a7 : 431 1029al9 : 433 I029a20 : 435 1029a21 : 434 1029a23 : 126, 434 1029a27 : 13, 42n 1029a29 : 43 1029a31 : 126 1029b3 : 230, 233 1029b!4 : 51, SSn 1029bl 6 : 370 n 1030a3 : 125 I030al0 : 124 1030al I : 1 4 n 1030al 4 : 125 I030al7 : 1 4 n 1030a22 : 1 7 1030b l l : 162 103la7 : 133 103la28 : 124 I03lbl3 : 124 103 l b14 : 126 103lb2S : 126 1032a6 : S9 n 1032aS : SS n 1032a28 : I02 1032a32 : 400 n 1032bl : 22n, 1 1 4 1032b2 : 24 n !032bl2 : 22n 1033a4 : 27 I033a30 : 122

523

1033a31 : 93 1033b!6 : 24n 1033b25 : 24 n !034aS : 50 I034b4 : I02 1034b29 : 1 4 I034b32 : 135 1035al : 26n 1035a2 : 135 !035a4 : 136, 137 1035a9 : 136 1035a!S : l l S 1035a21 : 136, 137, 139 1035a22 : 136 I035a23 : 13S 1035a33 : 14, 25n, 26n 1035bl : 142 1035b4 : !3S 1035bl2 : 22n, 13S I035bl4 : 25n, 26n, 1 1 4 1035bl5 : 139 I 035bl6 : 2S n 1035b22 : 437 1035b25 : 43S n 1035b26 : 14 1035b27 : 16, 27, 13S, 139, 142, 143, 143 n, 437 1035b28 : 24 n 1035b29 : 139 1035b32 : 1 1 4 1035b33 : l l S , 139 1035b34 : 26 1036al : 25n, 27n !036a2 : 26, 27 1036a5 : 27n 1036a7 : 27 n 1036al2 : 139 1036a28 : 26 1036a32 : 50 1036b3 : 14 n !036bl0 : 1 4 n, l l S 1036b21 : 2S 1036b26 : l !S, 1 1 9 1036b28 : 14, 1 4 n , 2Sn, 1 17, 121, 122 1036b29 : 1 19 1036b31 : 14 1037a5 : 437 I 037a6 : 24 n, 142, 143 1037a7 : 26n 1037a21 : 1 1 6 1037a24 : 25n, 27n, 1 16, 135, 137, !3S, 139 1037a25 : 27 n 1037a27 : 27n, 1 17, 123 1037a2S : 24 n

524

INDEX DES PASSAGES CITES

INDEX DES PASSAGES CITES

I037a30 : 43 1037a37 : 27 n I037bl l : 127 I037bl8 : 127 I037b25 : 37 I038a5 : 127 I038a6 : 439 I038a8 : 128 1038al 9 : 128 I038a25 : 128 I038a33 : 52 I038b23 : 439 I039a3 : 13, 439 I039al 4 : 129 I039b20 : 24n, 25n, 26n 1039b24 : 92 1039b27 : 26 I039b28 : 26 n I040a5 : 26 I040a8 : 26 n I040a27 : 441 n l 040b5 : IO n, 13, 13 n, 437, 440 n 1040b6 : 13, 44 I042al 2 : 44 I042a 16 : 45 I042a21 : 45 I042a26 : 24 n )042a32 : 43, 45 I042b9 : 433 I042bl 2 : 440n 1043a6 : 440 n I043al 4 : 93 1043b31 : 439 I043b36 : 441 n 1045a20 : 439 I045bl 7 : 440 I045bl 9 : 26n I045b23 : 26n I046b29 : 323 n 1047a l 0 : 323n I047a24 : 324 n 1047bl 4 : 324n 1047b39 : 231 I048a35 : 383 n I048b37 : 231 I049bl : 123 I050a2 : 344 I054b33 : 383n I055a6 : 383 n 1055bl7 : 383 n 1057b32 : 440n I058a9 : 383 n I058b6 : 498 I067bl 4 : 383 n

l 070a7 l 07 l a l 2 I073a31 I074b l I074b9 I075a l 9 l 075bl ! I077a27 I078bl 7 l 086b2 I087b33

: 396 n : 483 n : 478 : 473 n : 473 n : 489n : 473n : 502 : 263 n : 383 n, 386 n : 442 n

Ethique a Nicomaque I094a26 : 400n I095a28 : 2 1 1 I095b6 : 339 I098a2 : 413 I098b3 : 372 n 1098bl 2 : 220 I098h27 : 2 1 1 , 229 l l 0 la22 : 220 l l l l b6 : 408 n 1 139a l l : 407 n l 139al8 : 413 l 139a35 : 400 n l 139b27 : 374 n ! ! 39b28 : 371 n ! ! 4 l a9 : 209 l 1 4 l a22 : 421 n4la25 : 421 n l 1 4 la26 : 405 ! !44a l 4 : 407n 1 1 45b2 : 267n l l 45b8 : 267 n l 145b l 9 : 220 1 1 45b25 : 267n ! !46a4 : 208 n ! ! 46a5 : 224 l 146a6 : 220 ! ! 46b4 : 220 l 146b6 : 267 n ! ! 46b23 : 222 l l 46b35 : 223 l 147a9 : 223 ! !47al8 : 223 l 147b3 : 408 n ! ! 49b31 : 408 n l l 53a8 : 406 l l 72b35 : 208 n 1 1 76a3 : 422n Ethique a Eudeme 1216b26 : 229, 232, 278 n

Politiques 1256bl6 l 277b25 128 l b l 5 1281 b35 1282al 5 1332b3 1336b4

: : : : : : :

41 406 n 230 230 210 406 n 4!0

Economique 1343bl5 : 4 1 7 n Rhelorique 1354al : 187, 216, 279n, 281 n 1354a2 : 187, 207 1354a3 : 278 n 1354a4 : 187 1354a5 : 188, 213 1355al 4 : 229, 285n 1355a21 : 284 n

1355a24 1355bl 5 1356a25 1356b3 1 356b8 1358al 1358a2 1358al 0 1358al 2 1358a26 1359al 6 1359b9 l 393a32 1395b30

525

: 188, 217, 278 n, 279 : 277n : 284 n : 374 n, 376 n : 371 n : 217 n : 280 n : 279 n, 280n : 217n : 218n : 217n : 284 n : 373 n : 218n

Poetique 1459b41 : 62n Fragments

(ed. Ross)

Sophisle, frag. 1 : 263

!

!

TABLE DES MATIERES

D . DEVEREUX, P. PELLEGRIN

:

introduction . . . . . . . . . . . . . . . . .

.

G. E . R . LLOYD : Aristotle's zoology and his metaphysics. The status quaestionis. A critical review of some recent theories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

7

P . PELLEGRIN : Taxinomie, moriologie, division : reponses a G. E . R . LLoyd . . . . . . . . . . . . . . . . . . . . . : . . . .

37

D . M . BALME : Matter in definition : a reply to G. E . R. Lloyd . .

.

49

J . M . Coo PER

Metaphysics in Aristotle's embryology . . . . . . . .

.

55

M . FURTH : Specific and individual form in Aristotle . . . . . . . . .

.

85

.

.

:

I,:

.

.

.

.

.

.

.

.

.

.

M . FREDE : The definition of sensible substances in Metaphysics Z . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 13

D . MORRISON : Some remarks on definition in Metaphysics Z . . .

.

130

D . CHARLES : Aristotle on meaning, natural kinds and natural history J . G. LENNOX : Notes on David Charles on H.A

.

J'

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

1 45

.

.

.

.

.

.

.

.

.

. ....

1 69

.

.

.

.

.

R. BOLTON : The epistemological basis of Aristotelian dialectic . .

.

185

J . BRUNSCHWIG : Remarques sur la communication de Robert 237 Bolton . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

D . DEVEREUX : Comments on Robert Bolton's "The epistemo263 logical basis of Aristotelian dialectic" . . . . . . . . . . . . . . . . . .

C. FREELAND : Scientific explanation and empirical data in 287 Aristotle's Meteorology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

M . M1GNucc1 : Aristotle's De Caelo I , 1 3 and his notion of possibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321 W. KULLMANN : Bipartite science in Aristotle's biology . . . . . . . . 335 A. KosMAN : Necessity and explanation in Aristotle's Analylics . . .

349

v

v�--

,__

--r.....---

\

I/

{

/

TABLE DES MATIERES

528

F. CAUJOLLE-ZASLAWSKY : Etude preparatoire a une interpretation du sens aristotelicien d'btocywyij . . . . . . . . . . . . . . . . . 365 S. W. BROADIE : Nature and craft in Aristotelian teleology . . . , . 389 J . L . LABARRIERE : De la phronesis animale . . . . . . . . . . . . . . . . . . 405 T. MAUDLIN : Substance and Space-Time: what Aristotle would have said to Einstein . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429 A . PREUS : Man and cosmos in Aristotle: Metaphysics A and the biological works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471 R. THOM : Homeomeres et anhomeomeres en theorie biologique d'Aristote a aujourd'hui . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 491 INDEX DES PASSAGES D1ARISTOTE CITES . . . , , . . . . . . . . . . , , . . . . ,

IMPRIMERIE

A.

BONTEMPS, LIMOGES.

-

Depot legal

:

Novembre 1990. - N° IMP.

:

513

1593-90